SEARCH SESSIONS

Search All
Refine Results:
 
Year(s)

SOCIAL MEDIA

EMAIL SUBSCRIPTION

 
 

GTC ON-DEMAND

Deep Learning and AI Frameworks
Presentation
Media
Opening Keynote (Keynote Talk)
Abstract:
Don't miss this keynote from NVIDIA Founder & CEO, Jensen Huang, as he speaks on the future of computing.
 
Topics:
Deep Learning and AI Frameworks, Robotics, Intelligent Machines and IoT, Autonomous Vehicles, Data Center and Cloud Computing
Type:
Keynote
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9688
Streaming:
Share:
AI Application Deployment and Inference
Presentation
Media
Abstract:
This talk addresses how enterprises can reduce risk by taking a practical approach with their AI initiatives from conceptualization to deployment by evaluating business value, managing open source effectively, creating development environments that e ...Read More
Abstract:
This talk addresses how enterprises can reduce risk by taking a practical approach with their AI initiatives from conceptualization to deployment by evaluating business value, managing open source effectively, creating development environments that economically scale over time as the needs grow, and creating effective workflows based on data movement. By taking this approach to AI, enterprises customers can make incremental investments for success while overcoming the complexity of this new technology  Back
 
Topics:
AI Application Deployment and Inference, AI and DL Business Track (high level)
Type:
Sponsored Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S91000
Streaming:
Share:
 
Abstract:
Advancements in the field of artificial intelligence, particularly in the area of deep learning, have facilitated training and inferencing to bring new applications to every aspect of our lives. This session is designed to bridge the gap by identifyi ...Read More
Abstract:
Advancements in the field of artificial intelligence, particularly in the area of deep learning, have facilitated training and inferencing to bring new applications to every aspect of our lives. This session is designed to bridge the gap by identifying what is required to apply AI to a range of new types of workloads.  Back
 
Topics:
AI Application Deployment and Inference
Type:
Sponsored Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S91007
Streaming:
Download:
Share:
 
Abstract:
Modern AI methods represent a great opportunity to rethink every aspect of our companies, from engineering products, to pricing them, to production, and even to basic tasks like human resources and finance. In this talk, we will talk about some use c ...Read More
Abstract:
Modern AI methods represent a great opportunity to rethink every aspect of our companies, from engineering products, to pricing them, to production, and even to basic tasks like human resources and finance. In this talk, we will talk about some use cases in the enterprise, and then the software tools and hardware infrastructure required to build successful AI-based applications. We will talk about some of ways IBM is enhancing open-source software like Tensorflow and pyTorch and also how automatic AI (AutoAI) will enable faster creation and deployment of AI models.  Back
 
Topics:
AI Application Deployment and Inference, AI and DL Business Track (high level)
Type:
Sponsored Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S91033
Streaming:
Download:
Share:
 
Abstract:
Exploring the Best Server for AI Speaker: Samuel D. Matzek, Sr. Software Engineer Speaker: Maria Ward, IBM Accelerated Server Offering Manager Explore the server at the heart of the Summit and Sierra supercomputers, and the best server for ...Read More
Abstract:

Exploring the Best Server for AI Speaker: Samuel D. Matzek, Sr. Software Engineer Speaker: Maria Ward, IBM Accelerated Server Offering Manager Explore the server at the heart of the Summit and Sierra supercomputers, and the best server for AI. We will discuss the technical details that set this server apart and why it matters for your machine learning and deep learning workloads. IBM Cloud for AI at Scale Speaker: Alex Hudak, IBM Cloud Offering Manager AI is fast changing the modern enterprise with new applications that are resource demanding, but provide new capabilities to drive insight from customer data. IBM Cloud is partnering with NVIDIA to provide a world class and customized cloud environment to meet the needs of these new applications. Learn about the wide range of NVIDIA GPU solutions inside the IBM Cloud virtual and bare metal server portfolio, and how customers are using them across Deep Learning, Analytics, HPC workloads, and more. IBM Spectrum LSF Family Overview & GPU Support Speaker: Larry Adams, Global Architect - Cross Sector, Developer, Consultant, IBM Systems How to Fuel the Data Pipeline Speaker: Kent Koeninger, IBM IBM Storage Reference Architecture for AI with Autonomous Driving Speaker: Kent Koeninger, IBM  

  Back
 
Topics:
AI Application Deployment and Inference
Type:
Sponsored Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S91053
Streaming:
Download:
Share:
 
Abstract:
Learn how recent achievements in machine learning, sensor fusion, and GPU computing make it possible to create a next-generation advanced driver-assistance systems experience. We'll showcase a software solution that creates real-time augmented reali ...Read More
Abstract:
Learn how recent achievements in machine learning, sensor fusion, and GPU computing make it possible to create a next-generation advanced driver-assistance systems experience. We'll showcase a software solution that creates real-time augmented reality for drivers while using vehicle sensors, map data, telematics, and navigation guidance with a set of advanced algorithms. Our approach augments drivers' visual reality with supplementary objects in real time, and works with output devices such as head unit displays, digital clusters, and head-up displays. We'll also examine the challenges of running advanced neural network models in real time on embedded hardware and explain solutions to overcome them.  Back
 
Topics:
AI Application Deployment and Inference, Virtual Reality and Augmented Reality, Autonomous Vehicles
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9169
Streaming:
Download:
Share:
 
Abstract:
A common deep learning workload is batch processing of videos to identify objects an image. We'll show examples of how to deploy a style-transfer and object-detection model on a cluster of V100 GPUs using Dask. Dask allows us develop the logic of ou ...Read More
Abstract:
A common deep learning workload is batch processing of videos to identify objects an image. We'll show examples of how to deploy a style-transfer and object-detection model on a cluster of V100 GPUs using Dask. Dask allows us develop the logic of our processing pipeline locally and deploy it on a cluster without having to rewrite anything. We'll discuss how we integrate it into Azure ML pipelines, as well as how to deploy it on a Kubernetes cluster for a scalable solution.  Back
 
Topics:
AI Application Deployment and Inference, Deep Learning and AI Frameworks
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9198
Streaming:
Download:
Share:
 
Abstract:
We'll present a fast, highly accurate, and customizable object-detection network optimized for training and inference on GPUs. After describing the network architecture, we'll dive into how different stages of training workflow are accelerated. Our ...Read More
Abstract:
We'll present a fast, highly accurate, and customizable object-detection network optimized for training and inference on GPUs. After describing the network architecture, we'll dive into how different stages of training workflow are accelerated. Our techniques include data ingestion and augmentation, mixed precision, and multi-GPU training. We'll demonstrate how we optimized our network for deployment without loss of accuracy using ONNX and NVIDIA TensorRT. We'll also show how to create TensorRT plugins for post-processing to perform inference entirely on the GPU. This session will be a combination of lecture and demos.  Back
 
Topics:
AI Application Deployment and Inference, Deep Learning and AI Frameworks, Computer Vision
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9243
Streaming:
Download:
Share:
 
Abstract:
We'll discuss our work at Esri to reconstruct 3D building models from aerial LiDAR data with the help of deep neural networks. The value of accurate 3D building models for cities is hard to overestimate, but collecting and maintaining this data is l ...Read More
Abstract:
We'll discuss our work at Esri to reconstruct 3D building models from aerial LiDAR data with the help of deep neural networks. The value of accurate 3D building models for cities is hard to overestimate, but collecting and maintaining this data is labor-intensive, error-prone, and expensive. We teamed up with Miami-Dade County and NVIDIA to see if we could streamline this data-acquisition workflow or at least, make it more cost-effective. We used a Mask R-CNN model trained to detect and report instances of roof segments of various types. Our talk will cover data preparation and Mask R-CNN training and achieved precision. We'll also outline the inference architecture, the integration of TensorFlow and ArcGIS Pro 2.3, and the steps we used to reconstruct 3D building models from the predictions.  Back
 
Topics:
AI Application Deployment and Inference, Seismic and Geosciences, Product & Building Design
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9255
Streaming:
Download:
Share:
 
Abstract:
Take a journey through the TensorFlow container provided by the NVIDIA GPU Cloud. We'll start with how to launch and navigate inside the container, and stop along the way to explore the included demo scripts, extend the container with extra ...Read More
Abstract:

Take a journey through the TensorFlow container provided by the NVIDIA GPU Cloud. We'll start with how to launch and navigate inside the container, and stop along the way to explore the included demo scripts, extend the container with extra software, and examine best practices for how to take advantage of all the benefits bundled inside the NGC TensorFlow container. This session will help NGC beginners get the most out of the TensorFlow container and become productive as quickly as possible.

  Back
 
Topics:
AI Application Deployment and Inference, Deep Learning and AI Frameworks
Type:
Tutorial
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9256
Download:
Share:
 
Abstract:
Learn how GPU Coder produces high-performance CUDA code automatically from a high-level algorithm description in MATLAB. Write your deep learning application with the expressive power of MATLAB, which allows you to describe not just the use of your t ...Read More
Abstract:
Learn how GPU Coder produces high-performance CUDA code automatically from a high-level algorithm description in MATLAB. Write your deep learning application with the expressive power of MATLAB, which allows you to describe not just the use of your trained deep learning model in inference mode, but also perform data-augmentation and post-processing of the results to create a complete deployment-ready application. With MATLAB running on your host machine, communicate and control peripheral devices on your Jetson Xavier and DRIVE Xavier platforms to bring in live data from sensors for visualization and analysis. GPU Coder can then generate optimized inference code for the whole application. The deep learning inference model is compiled down to TensorRT's inference engine, while the rest of the application logic is parallelized through creation of CUDA kernels and integrated with other CUDA optimized libraries like cuBLAS, cuFFT, etc. GPU Coder provides a clean, elegant solution to go from algorithm to application deployment, unleashing the performance of CUDA, TensorRT, and the Xavier device architecture.  Back
 
Topics:
AI Application Deployment and Inference, Deep Learning and AI Frameworks, Computer Vision
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9281
Streaming:
Download:
Share:
 
Abstract:
Learn how to develop faster, scalable, and better GPU-Accelerated distributed inference on multi-node and multi-GPU cluster environments. In most of the benchmark cases, linear scalability for throughput performance is not guaranteed with increasing ...Read More
Abstract:
Learn how to develop faster, scalable, and better GPU-Accelerated distributed inference on multi-node and multi-GPU cluster environments. In most of the benchmark cases, linear scalability for throughput performance is not guaranteed with increasing the number of GPUs and servers. We'll discuss present an efficient scale-out method for deploying deep learning-based object detection models on multi-node and multi-GPU clusters using Apache Hadoop and Spark. We'll explain how to deploy 120 deep learning models (YOLOv2) on our own video datasets with NVIDIA Tesla M60 GPU (30EA) Hadoop cluster. Our approach achieved about 20 percent faster inference throughput with super-linear scalability from one GPU server to 30 GPU cluster. This session will be a combination of lecture and videos about our GPU-Accelerated distributed inference platform for large-scale streaming data analytics.  Back
 
Topics:
AI Application Deployment and Inference, Intelligent Video Analytics, Video and Image Processing
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9343
Streaming:
Download:
Share:
 
Abstract:
We'll talk about how we built a GPU-Accelerated system for real-time information retrieval from large datasets in life sciences. Unstructured textual data is full of phrases and words that have multiple meanings, making it difficult for current info ...Read More
Abstract:
We'll talk about how we built a GPU-Accelerated system for real-time information retrieval from large datasets in life sciences. Unstructured textual data is full of phrases and words that have multiple meanings, making it difficult for current information-retrieval algorithms to find relevant documents. We'll describe our knowledge graph-based filtering mechanism for more precise real-time information retrieval. We outline how we accelerated the embedding generation process, treating it as an optimization problem and running it on NVIDIA Tesla V100 GPU cores. We'll also cover how we reduced the latency in distance computation using TensorRT.  Back
 
Topics:
AI Application Deployment and Inference, Speech and Language Processing
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9359
Streaming:
Download:
Share:
 
Abstract:
Learn how recent advances in AI can be used to map informal settlements, or slums, in developing countries. We'll show how slums can be mapped using machine learning with noisy annotations and multi-resolution, multi-spectral data. We'll discuss an ...Read More
Abstract:
Learn how recent advances in AI can be used to map informal settlements, or slums, in developing countries. We'll show how slums can be mapped using machine learning with noisy annotations and multi-resolution, multi-spectral data. We'll discuss an effective end-to-end framework that detects and maps the locations of informal settlements using low-resolution, freely available Sentinel-2 satellite imagery. Our talk will examine different approaches based on multi-spectral information to identify roofing material types and show how our work can be used for slums all over the world. We'll also describe how multi-spectral, multi-resolution and multi-temporal satellite imagery can be used during natural disasters to quantify the impact on urban infrastructure. This session presents research undertaken in the NASA and ESA Frontier Development Lab.  Back
 
Topics:
AI Application Deployment and Inference, AI and DL Research, Computer Vision
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9362
Streaming:
Download:
Share:
 
Abstract:
We'll describe our work at Intelligent Voice on explainable AI. We are working to separate AI technology into smaller components so it can be more easily explained, build explainability into AI architecture design, and make it possible for A ...Read More
Abstract:

We'll describe our work at Intelligent Voice on explainable AI. We are working to separate AI technology into smaller components so it can be more easily explained, build explainability into AI architecture design, and make it possible for AI to progress within confines of current regulation. New GDPR regulations in Europe, which affect any company with European consumers, give people a right to challenge computer-aided decisions and to have these decisions explained. We'll discuss how existing technology can make it difficult to provide an explanation and how that inhibits AI adoption in customer-facing fields such as insurance, health, and financial services.

  Back
 
Topics:
AI Application Deployment and Inference, Advanced AI Learning Techniques
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9392
Streaming:
Download:
Share:
 
Abstract:
We will describe a new API that more effectively utilizes the GPU hardware for multiple single inference instances of the same RNN model. Many NLP applications have real-time run time requirements for multiple independent inference instances. Our pro ...Read More
Abstract:
We will describe a new API that more effectively utilizes the GPU hardware for multiple single inference instances of the same RNN model. Many NLP applications have real-time run time requirements for multiple independent inference instances. Our proposed API accepts independent inference requests from an application and seamlessly combines them to a large batch execution. Time steps from independent inference tasks are combined together so that we achieve high performance while staying within the latency budgets of an application for a time step. We also discuss functionality that allows the user to wait on completion of a certain time step, a task that's possible because our implementation is mainly composed of non-blocking function calls. Finally, we'll present performance data from the Turing architecture for an example RNN model with LSTM cells and projections.  Back
 
Topics:
AI Application Deployment and Inference
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9422
Streaming:
Download:
Share:
 
Abstract:
We'll explain how to use TensorRT via TensorFlow and/or TensorFlow serving. TensorFlow is a flexible, high-performance software library for numerical computation using data flow graphs and NVIDIA TensorRT is a platform for high-performance deep lear ...Read More
Abstract:
We'll explain how to use TensorRT via TensorFlow and/or TensorFlow serving. TensorFlow is a flexible, high-performance software library for numerical computation using data flow graphs and NVIDIA TensorRT is a platform for high-performance deep learning inference. We'll describe how TensorRT is integrated with TensorFlow and show how combining the two improves efficiency of machine learning models. We'll also use examples to show how to use the integration.  Back
 
Topics:
AI Application Deployment and Inference
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9431
Streaming:
Download:
Share:
 
Abstract:
As the use of AI has increased, so has the need for a production-quality AI inference solution. We'll discuss the latest additions to NVIDIA's TensorRT Inference Server and describe deployment examples to help plan your data center production infer ...Read More
Abstract:
As the use of AI has increased, so has the need for a production-quality AI inference solution. We'll discuss the latest additions to NVIDIA's TensorRT Inference Server and describe deployment examples to help plan your data center production inference architecture. NVIDIA TensorRT Inference Server makes it possible to efficiently leverage inference in applications and to do so without reinventing the wheel. We'll talk about how TensorRT supports the top AI frameworks and custom backends, and maximizes utilization by hosting multiple models per GPU and across GPUs with dynamic request batching. Our talk will also cover how the inference server seamlessly supports Kubernetes with health and latency metrics and integrates with Kubeflow for simplified deployment.   Back
 
Topics:
AI Application Deployment and Inference, Data Center and Cloud Infrastructure
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9438
Streaming:
Download:
Share:
 
Abstract:
Building machine learning pipelines is challenging. Doing that in a portable way that supports multi-cloud deployments is even harder. We'll discuss the open source project, Kubeflow, which is designed to allow data scientists and machine learning e ...Read More
Abstract:
Building machine learning pipelines is challenging. Doing that in a portable way that supports multi-cloud deployments is even harder. We'll discuss the open source project, Kubeflow, which is designed to allow data scientists and machine learning engineers to focus on building great ML solutions instead of setting up and managing the infrastructure. We'll detail the latest version of Kubeflow and its integration with TensorRT, the inference server from NVIDIA.  Back
 
Topics:
AI Application Deployment and Inference, Data Center and Cloud Infrastructure, Tools and Libraries
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9456
Streaming:
Download:
Share:
 
Abstract:
We'll explore the practical challenges involved in developing a deep learning and artificial intelligence platform for satellite imagery analysis. Our talk will cover technical challenges we faced, platform design considerations, and best practices ...Read More
Abstract:
We'll explore the practical challenges involved in developing a deep learning and artificial intelligence platform for satellite imagery analysis. Our talk will cover technical challenges we faced, platform design considerations, and best practices learned from implementing our analytics platform. Learn how to modify a deep learning model for better throughput by combining TensorFlow and TensorRT with the right hardware and software platform options. We'll explain how to design architecture to inference huge amounts of imagery using TensorRT and discuss business opportunities for the defense and financial industries that arise from watching whole world with a global dataset.  Back
 
Topics:
AI Application Deployment and Inference, Seismic and Geosciences
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9459
Download:
Share:
 
Abstract:
Kubeflow has emerged as the de facto way to do ML on Kubernetes. Learn about the fastest way to go from GPU machines to an operational Kubeflow cluster on any public or private cloud, bare metal, or even your own laptop. We'll demonstrate the tools ...Read More
Abstract:
Kubeflow has emerged as the de facto way to do ML on Kubernetes. Learn about the fastest way to go from GPU machines to an operational Kubeflow cluster on any public or private cloud, bare metal, or even your own laptop. We'll demonstrate the tools and steps necessary to quickly deploy Kubeflow at any scale. We'll also show you a fast and exciting new way to deploy Kubeflow on a single laptop or desktop, which is perfect for local ML development. Don't lose time deploying and configuring Kubernetes and Kubeflow get to the ML as quickly as possible.  Back
 
Topics:
AI Application Deployment and Inference, Deep Learning and AI Frameworks
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9515
Streaming:
Download:
Share:
 
Abstract:
Learn about the advantages and pitfalls of venturing away from off-the-shelf libraries to implement neural network inference algorithms from the ground up. We'll discuss the challenges of building large-vocabulary speech recognition engines able to ...Read More
Abstract:
Learn about the advantages and pitfalls of venturing away from off-the-shelf libraries to implement neural network inference algorithms from the ground up. We'll discuss the challenges of building large-vocabulary speech recognition engines able to support decoding more than 1,000 simultaneous conversations per NVIDIA V100 card, while still able to down-port onto low-memory embedded configurations such as the Tegra TK1. We'll cover what characteristics of the many types of popular neural networks used in speech recognition scale almost perfectly, as well as those that resist scaling and even scale negatively. Learn what profiling reveals about the silent, looming cost of kernel synchronization and what to do about it.  Back
 
Topics:
AI Application Deployment and Inference, Speech and Language Processing
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9535
Streaming:
Download:
Share:
 
Abstract:
We'll share our experience building an audio cognition platform that extracts non-verbal information such as speech, music, and environmental sounds. Cochlear.ai's Sense platform leverages acoustic event detection, scene classification, human gende ...Read More
Abstract:
We'll share our experience building an audio cognition platform that extracts non-verbal information such as speech, music, and environmental sounds. Cochlear.ai's Sense platform leverages acoustic event detection, scene classification, human gender/age estimation, music analysis, and more with near real-time analysis from audio data. Everything is optimized for audio processing, including the cloud backend, API design and management, and deep learning architecture. We'll detail our learnings and challenges in developing our audio cognition service.  Back
 
Topics:
AI Application Deployment and Inference, AI and DL Research, Deep Learning and AI Frameworks
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9625
Streaming:
Download:
Share:
 
Abstract:
You may already use NVIDIA's cuDNN library to accelerate your deep neural network inference, but are you getting the most out of it to truly unleash the tremendous performance of NVIDIA's newest GPU architectures, Volta and Turing? We'll discuss h ...Read More
Abstract:
You may already use NVIDIA's cuDNN library to accelerate your deep neural network inference, but are you getting the most out of it to truly unleash the tremendous performance of NVIDIA's newest GPU architectures, Volta and Turing? We'll discuss how to avoid the most common pitfalls in porting your CPU-based inference to the GPU and demonstrate best practices in a step-by-step optimization of an example network. Learn how to deploy your deep neural network inference in both the fastest and most memory-efficient way, using cuDNN and Tensor Cores, NVIDIA's revolutionary technology that delivers groundbreaking performance in FP16, INT8 and INT4 inference on Volta and Turing.  Back
 
Topics:
AI Application Deployment and Inference, Deep Learning and AI Frameworks
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9644
Streaming:
Download:
Share:
 
Abstract:
Learn how to bring the power of convolutional neural networks and deep learning to memory- and power-constrained devices like smartphones, wearable devices, and drones. We'll show these techniques at work on real-world project and discuss tips and t ...Read More
Abstract:
Learn how to bring the power of convolutional neural networks and deep learning to memory- and power-constrained devices like smartphones, wearable devices, and drones. We'll show these techniques at work on real-world project and discuss tips and tricks, speed and accuracy trade-offs, and benchmarks on different hardware. We will then demonstrate how to get started developing your own deep learning application for storage- and power-constrained mobile devices. We'll also discuss how to apply similar techniques to increase deep neural net efficiency when deploying in a regular cloud-based production environment. This approach reduces the number of GPUs required and lowers cost.  Back
 
Topics:
AI Application Deployment and Inference, AI and DL Research
Type:
Tutorial
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9648
Streaming:
Download:
Share:
 
Abstract:
We'll discuss Project MagLev, NVIDIA's internal end-to-end AI platform for developing its self-driving car software, DRIVE. We'll explore the platform that supports continuous data ingest from multiple cars producing TB of data per h ...Read More
Abstract:

We'll discuss Project MagLev, NVIDIA's internal end-to-end AI platform for developing its self-driving car software, DRIVE. We'll explore the platform that supports continuous data ingest from multiple cars producing TB of data per hour. We'll also cover how the platform enables autonomous AI designers to iterate training of new neural network designs across thousands of GPU systems and validate the behavior of these designs over multi PB-scale data sets. We will talk about our overall architecture for everything from data center deployment to AI pipeline automation, as well as large-scale AI dataset management, AI training, and testing.

  Back
 
Topics:
AI Application Deployment and Inference, Autonomous Vehicles, Data Center and Cloud Infrastructure
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9649
Streaming:
Download:
Share:
 
Speakers:
Abstract:
Although neural network training is typically done in either 32- or 16-bit floating point formats, inference can be run at even lower precisions that reduce memory footprint and elapsed time. We'll describe quantizing neural networks models for vari ...Read More
Abstract:
Although neural network training is typically done in either 32- or 16-bit floating point formats, inference can be run at even lower precisions that reduce memory footprint and elapsed time. We'll describe quantizing neural networks models for various image (classification, detection, segmentation) and natural language processing tasks. In addition to convolutional feed forward networks, we will cover quantization of recurrent models. The discussion will examine both floating point and integer quantizations, targeting features in Volta and Turing GPUs.  Back
 
Topics:
AI Application Deployment and Inference, AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9659
Streaming:
Download:
Share:
 
Abstract:
There are numerous problems which have been exposed by the creation of AI models due to the total capability of the current generation of GPUs to create and run a large volume of models, and we are going to show people how to fix them. The exponentia ...Read More
Abstract:
There are numerous problems which have been exposed by the creation of AI models due to the total capability of the current generation of GPUs to create and run a large volume of models, and we are going to show people how to fix them. The exponential compute growth which has occurred in this area has opened the doors to creating and testing hundreds or thousands more models than the, one-by-one approach which was performed in the past. These models use and generate data from both batch and real-time sources. As data becomes enriched, and parameters tuned and explored, there is a need for versioning everything, including the data. Issues found here are similar to other software engineering problems, but new approaches must be taken to create solutions given the complexity of the problems with the inclusion of vast amounts of data. We will discuss the very specific problems and approaches to fix them.  Back
 
Topics:
AI Application Deployment and Inference, Data Center and Cloud Infrastructure
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9683
Streaming:
Download:
Share:
 
Abstract:
We'll discuss network quantization its background, methods, achievements, and the motivation behind it. Deep neural networks have achieved remarkable performance in a wide range of tasks. But DNNs are computationally intensive and resource-consuming ...Read More
Abstract:
We'll discuss network quantization its background, methods, achievements, and the motivation behind it. Deep neural networks have achieved remarkable performance in a wide range of tasks. But DNNs are computationally intensive and resource-consuming, which hinders their use in embedded systems. We'll explain how we're working to alleviate this problem with quantized neural networks and a lightweight framework for efficient inference of these networks.  Back
 
Topics:
AI Application Deployment and Inference
Type:
Tutorial
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9713
Streaming:
Download:
Share:
 
Abstract:
Many computer vision applications powered by deep learning include multi-stage pre-processing data pipelines with compute-intensive processes like decoding, cropping, and format conversion that are carried out on CPUs. We'll discuss NVIDIA DALI, an ...Read More
Abstract:
Many computer vision applications powered by deep learning include multi-stage pre-processing data pipelines with compute-intensive processes like decoding, cropping, and format conversion that are carried out on CPUs. We'll discuss NVIDIA DALI, an open source, GPU-Accelerated data augmentation and image-loading library for optimizing data pipelines of deep learning frameworks. DALI provides a full pre- and post-processing data pipeline ready for training and inference. We'll demonstrate a TensorRT inference workflow within DALI-configurable graphs as well as custom operators.  Back
 
Topics:
AI Application Deployment and Inference, Deep Learning and AI Frameworks
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9818
Streaming:
Download:
Share:
 
Abstract:
We'll demonstrate the use of Nsight Systems to quickly identify bottlenecks and achieve significant speedups in production workflows at Facebook. We'll also describe how we use the CUPTI API for on-demand, customized timeline analysis of workflows ...Read More
Abstract:
We'll demonstrate the use of Nsight Systems to quickly identify bottlenecks and achieve significant speedups in production workflows at Facebook. We'll also describe how we use the CUPTI API for on-demand, customized timeline analysis of workflows running in production and collect detailed performance metrics across our GPU fleet at very low overhead.  Back
 
Topics:
AI Application Deployment and Inference
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9866
Streaming:
Download:
Share:
 
Abstract:
We'll discuss how human rights organizations can use AI to investigate human rights violations around the world. With a focus on deep learning and satellite imagery, we'll examine potential applications, including signature detection of arson, hous ...Read More
Abstract:
We'll discuss how human rights organizations can use AI to investigate human rights violations around the world. With a focus on deep learning and satellite imagery, we'll examine potential applications, including signature detection of arson, housing demolition, and indiscriminate air strikes. Our talk will also cover some technical challenges posed by the lack of training data and the complexity of detecting relevant abuse signatures, as well as consider the potential utility of synthetic training data 3D modeling.  Back
 
Topics:
AI Application Deployment and Inference
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9947
Download:
Share:
 
Abstract:
How do you accelerate innovation and deliver faster time-to-value for your AI initiative, while ensuring enterprise-grade security and high performance? How do you provide easy access to the tools and data your data science teams need for large-scale ...Read More
Abstract:
How do you accelerate innovation and deliver faster time-to-value for your AI initiative, while ensuring enterprise-grade security and high performance? How do you provide easy access to the tools and data your data science teams need for large-scale distributed ML/DL with greater agility for rapid prototyping and iteration? We'll discuss practical examples and lessons learned from GPU-Accelerated ML/DL use cases in financial services, healthcare, and other industries. Learn how to quickly deploy containerized multi-node environments for TensorFlow and other ML DL tools in a multi-tenant architecture with a shared pool of resources using GPUs.  Back
 
Topics:
AI Application Deployment and Inference, AI and DL Business Track (high level)
Type:
Sponsored Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9982
Streaming:
Download:
Share:
AI and DL Business Track (high level)
Presentation
Media
Abstract:
Hundreds of talks and competing events crammed into a few days can be daunting. Get an overview of GTC's programs and events and how to make best use of them from Greg Estes, NVIDIA's VP of developer programs. Addressing both first-timer ...Read More
Abstract:

Hundreds of talks and competing events crammed into a few days can be daunting. Get an overview of GTC's programs and events and how to make best use of them from Greg Estes, NVIDIA's VP of developer programs. Addressing both first-timers and returning alums, Greg will cover how to get the most from your time here, including can't-miss talks and never-before-seen tech demos. He'll also cover NVIDIA's resources for developers, startups, and larger organizations, as well as training courses and networking opportunities.

  Back
 
Topics:
AI and DL Business Track (high level)
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S91004
Streaming:
Download:
Share:
 
Abstract:
A fireside chat with U.S. Rep. Jerry McNerney (D-Calif.), co-chair of the congressional AI caucus, and Ned Finkle, VP of Govt. Affairs, NVIDIA. Artificial Intelligence has become a front-and-center issue for policymakers. Legislative propos ...Read More
Abstract:

A fireside chat with U.S. Rep. Jerry McNerney (D-Calif.), co-chair of the congressional AI caucus, and Ned Finkle, VP of Govt. Affairs, NVIDIA. Artificial Intelligence has become a front-and-center issue for policymakers. Legislative proposals to encourage AI development and head off possible harms are gaining traction, and the Administration is working to build a national strategy. This fireside chat will give enterprises and researchers a first-hand look at how key Members of Congress are approaching AI, as well as what policies they're advocating for and expect.

  Back
 
Topics:
AI and DL Business Track (high level)
Type:
Panel
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S91006
Streaming:
Share:
 
Abstract:
Innovation is critical to every enterprise, but rarely convenient. We'll explore how business leaders approach the dual responsibility of building business growth and efficiencies through AI innovation while navigating the change cycles that every e ...Read More
Abstract:
Innovation is critical to every enterprise, but rarely convenient. We'll explore how business leaders approach the dual responsibility of building business growth and efficiencies through AI innovation while navigating the change cycles that every enterprise experiences.  Back
 
Topics:
AI and DL Business Track (high level)
Type:
Panel
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S91015
Streaming:
Download:
Share:
 
Abstract:
As machine learning and deep learning techniques evolve into mainstream adoption, the architectural considerations for platforms that support large-scale production deployments of AI applications change significantly. How do you ensure IO bottlenecks ...Read More
Abstract:
As machine learning and deep learning techniques evolve into mainstream adoption, the architectural considerations for platforms that support large-scale production deployments of AI applications change significantly. How do you ensure IO bottlenecks are eliminated to keep your GPU-Powered AI rocket ship fueled with data? How do you address the issues of data gravity, data scaling, and data economics to support Petabyte-sized data sets? How do you simplify data management and minimize business risk and life cycle costs of large scale AI platforms? Well address these questions, discuss key business and architectural requirements for compute and storage, and explain how enterprises can achieve the maximum benefit from AI platforms that align with these requirements. Well also introduce the Dell, EMC, and NVIDIA solution portfolio which makes AI simple, flexible, and accessible.  Back
 
Topics:
AI and DL Business Track (high level), AI and DL Research
Type:
Sponsored Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S91016
Streaming:
Download:
Share:
 
Abstract:
This customer panel brings together AI implementers who have deployed deep learning at scale. The discussion will focus on specific technical challenges they faced, solution design considerations, and best practices learned from implementing the ...Read More
Abstract:

This customer panel brings together AI implementers who have deployed deep learning at scale. The discussion will focus on specific technical challenges they faced, solution design considerations, and best practices learned from implementing their respective solutions.

  Back
 
Topics:
AI and DL Business Track (high level), Data Center and Cloud Infrastructure, Deep Learning and AI Frameworks
Type:
Panel
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9121
Streaming:
Share:
 
Abstract:
The size and complexity of problems that can be tackled with machine learning, and particularly deep learning, represents a new approach to business problem-solving. Learn to identify use cases for ML and acquire best practices to frame problems in a ...Read More
Abstract:
The size and complexity of problems that can be tackled with machine learning, and particularly deep learning, represents a new approach to business problem-solving. Learn to identify use cases for ML and acquire best practices to frame problems in a way that key stakeholders and senior management can understand and support. In our talk, we'll help delegates set the stage for delivering successful ML-based solutions to their businesses.  Back
 
Topics:
AI and DL Business Track (high level), Accelerated Data Science
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9295
Streaming:
Download:
Share:
 
Abstract:
Designed for the business leader, this session is a getting started primer for deep learning in the enterprise. Through cross-industry use cases, panelists will discuss adoption considerations, developing teams, building proof-of-concepts, and measur ...Read More
Abstract:
Designed for the business leader, this session is a getting started primer for deep learning in the enterprise. Through cross-industry use cases, panelists will discuss adoption considerations, developing teams, building proof-of-concepts, and measurement.  Back
 
Topics:
AI and DL Business Track (high level)
Type:
Panel
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9937
Streaming:
Share:
 
Abstract:
Artificial Intelligence has the potential to profoundly affect our world and lives. In this era of constant change, how do organizations keep up? We'll discuss the forces that drive technology forward and the technology trends, including AI, ...Read More
Abstract:

Artificial Intelligence has the potential to profoundly affect our world and lives. In this era of constant change, how do organizations keep up? We'll discuss the forces that drive technology forward and the technology trends, including AI, that can help organizations remain relevant in a world of constant transformation.

  Back
 
Topics:
AI and DL Business Track (high level)
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9938
Streaming:
Share:
 
Abstract:
By 2035, artificial intelligence could increase productivity by 40 percent or more. Manufacturing, healthcare, retail, and other key industries will benefit. We'll discuss how we're driving operational efficiencies within our organizations with AI ...Read More
Abstract:
By 2035, artificial intelligence could increase productivity by 40 percent or more. Manufacturing, healthcare, retail, and other key industries will benefit. We'll discuss how we're driving operational efficiencies within our organizations with AI applications, from getting started to advanced systems.  Back
 
Topics:
AI and DL Business Track (high level)
Type:
Panel
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9941
Streaming:
Share:
 
Abstract:
Retailers are on the front lines of using AI as an instrument of innovation, ranging from in-store experiences to back-end efficiencies. We'll provide practical insights from retail industry leaders that outline success criteria that any industry ca ...Read More
Abstract:
Retailers are on the front lines of using AI as an instrument of innovation, ranging from in-store experiences to back-end efficiencies. We'll provide practical insights from retail industry leaders that outline success criteria that any industry can apply to its strategy, from optimization of supply chain routes to spend management.  Back
 
Topics:
AI and DL Business Track (high level)
Type:
Panel
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9942
Streaming:
Share:
 
Abstract:
Data quality is a challenge all businesses face as they move forward with AI applications. Real-world data can be flawed, and attempting to create it for training and inference is sometimes too dangerous or costly to tackle. We'll discuss synthetic ...Read More
Abstract:
Data quality is a challenge all businesses face as they move forward with AI applications. Real-world data can be flawed, and attempting to create it for training and inference is sometimes too dangerous or costly to tackle. We'll discuss synthetic data, which can be created in controlled environments, even virtual ones. We'll explain why we see it as the future of data generation that will help businesses succeed.  Back
 
Topics:
AI and DL Business Track (high level)
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9943
Streaming:
Download:
Share:
 
Abstract:
How has AI had an effect on the world around us? Hear the extraordinary stories showcased in the I am AI keynote video and documentary series, and learn how humans and technology are working together to solve the grand challenges of our time. ...Read More
Abstract:
How has AI had an effect on the world around us? Hear the extraordinary stories showcased in the I am AI keynote video and documentary series, and learn how humans and technology are working together to solve the grand challenges of our time.  Back
 
Topics:
AI and DL Business Track (high level)
Type:
Panel
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9944
Streaming:
Share:
 
Abstract:
Healthcare has been on the frontlines of AI adoption for some time, using it to solve critical problems impacting humanity at large. What can other industries learn from healthcare AI innovation, in terms of applications and benchmarks? What insights ...Read More
Abstract:
Healthcare has been on the frontlines of AI adoption for some time, using it to solve critical problems impacting humanity at large. What can other industries learn from healthcare AI innovation, in terms of applications and benchmarks? What insights can be shared about cross-industry concerns such as privacy and security? How will healthcare further evolve now that data-centric models are driving new models?  Back
 
Topics:
AI and DL Business Track (high level)
Type:
Panel
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9989
Streaming:
Share:
AI and DL Research
Presentation
Media
Abstract:
Learn how to design GPU-Based systems for different application scenarios. We'll explain how to design data centers for different scales, application scenarios, and standards for enterprises and hyperscalers. We'll cover AI training and inference a ...Read More
Abstract:
Learn how to design GPU-Based systems for different application scenarios. We'll explain how to design data centers for different scales, application scenarios, and standards for enterprises and hyperscalers. We'll cover AI training and inference applications and edge computing for OCP and ODCC standard data centers. We'll discuss the challenges involved and share our experience designing a GPU platform for data centers. We'll also explore problems attendees are facing and see how we can work together to solve them.  Back
 
Topics:
AI and DL Research, HPC and AI
Type:
Sponsored Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S91013
Streaming:
Share:
 
Abstract:
Learn about the latest research on improvements to text-to-speech models and workflows using Tacotron2 and Waveglow produced by NVIDIA's applied deep learning research team. In partnership with our deep learning algorithm development team, learn mor ...Read More
Abstract:
Learn about the latest research on improvements to text-to-speech models and workflows using Tacotron2 and Waveglow produced by NVIDIA's applied deep learning research team. In partnership with our deep learning algorithm development team, learn more about how Tensor Cores have made fast mixed-precision training and faster than real-time inference performance available. We'll also be showing a demo, reviewing accuracy, and performance metrics through the open source implementation available on GitHub.  Back
 
Topics:
AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S91022
Streaming:
Download:
Share:
 
Abstract:
In order to obtain peak performance and energy efficiency on modern deep learning architectures, such as GPUs and TPUs, it is critical to use half precision arithmetic. Compared to single precision, half precision reduces memory traffic, allowing 2x ...Read More
Abstract:
In order to obtain peak performance and energy efficiency on modern deep learning architectures, such as GPUs and TPUs, it is critical to use half precision arithmetic. Compared to single precision, half precision reduces memory traffic, allowing 2x better use of the available DRAM bandwidth. Smaller memory footprints for half precision layer activations also allow larger batch sizes and deeper network architectures to fit in the accelerator's memory during training. Finally, architectural features, such as Volta's Tensor Cores, boost the raw math throughput of half precision operations by up to 8x compared to single precision. We describe two new streamlined implementations of mixed-precision training being built into TensorFlow. The first is provided through extensions to the tf.keras API and will be available in the upcoming months. The second is based on a Grappler graph optimization pass and will work with TF 1.x graph-based models as well as future TensorFlow 2.0 models that make use of tf.function decorators. Each method is enabled using a one or two line tweak to the training script. Empirical results show that result accuracy matches that of a model trained in single-precision, while training speedup is similar to what can be achieved with hand-coded mixed precision strategies.  Back
 
Topics:
AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S91029
Streaming:
Download:
Share:
 
Abstract:
GPUs are powering us into the future. Through research, we will be able to see where this technology takes us next. This session serves as a catalyst for the advancement of an array of innovations that come from universities, research labs, an ...Read More
Abstract:
GPUs are powering us into the future. Through research, we will be able to see where this technology takes us next. This session serves as a catalyst for the advancement of an array of innovations that come from universities, research labs, and industry. Join us in hearing how this year’s Top 5 Poster finalists were inspired to start their research, what surprised them, and where they want to see it go next.
 
  Back
 
Topics:
AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S91036
Streaming:
Download:
Share:
 
Abstract:
I will introduce a game developed at Johns Hopkins University/Applied Physics Laboratory called Reconnaissance Blind Chess (RBC), a chess variant where the players do not see their opponent's moves, but they can gain information about t ...Read More
Abstract:

I will introduce a game developed at Johns Hopkins University/Applied Physics Laboratory called Reconnaissance Blind Chess (RBC), a chess variant where the players do not see their opponent's moves, but they can gain information about the ground-truth board position through the use of an (imperfect) sensor.  RBC incorporates key aspects of active sensing and planning: players have to decide where to sense, use the information gained through sensing to update their board estimates, and use that world model to decide where to move.  Thus, just as chess and go have been challenge problems for decision making with complete information, RBC is intended to be a common challenge problem for decision making under uncertainty.  After motivating the game concept and its relationship to other chess variants, I will describe the current rules of RBC as well as other potential rulesets, give a short introduction to the game implementation and bot API, and discuss some of our initial research on the complexity of RBC as well as bot algorithm

  Back
 
Topics:
AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S91044
Streaming:
Download:
Share:
 
Abstract:
We'll discuss how using neural networks and deep reinforcement learning can be used to design potential drug candidates. The pharmaceutical industry is crying out for a revolution in thinking and practice; the traditional methods of drug discovery a ...Read More
Abstract:
We'll discuss how using neural networks and deep reinforcement learning can be used to design potential drug candidates. The pharmaceutical industry is crying out for a revolution in thinking and practice; the traditional methods of drug discovery and development are no longer working well. To continue to prosper, either R&D costs must be lowered or the rate of discovery for the new drugs must be drastically increased. We'll talk about how AI offers an opportunity to transform the field and dramatically accelerate the design of new drug candidates. The unique proposition of AI is the ability to learn directly from past experience and capture hidden dependences from both structured and unstructured data. As the chemical data is getting bigger, deep learning methods coupled with fast GPU computations make it possible to process vast amounts of information to find clinically relevant relationships and overcome drug discovery bottlenecks.  Back
 
Topics:
AI and DL Research, Computational Biology and Chemistry, AI in Healthcare
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9110
Streaming:
Download:
Share:
 
Abstract:
We'll introduce Data2Vis, a neural translation model for automatically generating visualizations from given datasets. We formulate visualization generation as a sequence-to-sequence translation problem in which data is mapped to visualization specif ...Read More
Abstract:
We'll introduce Data2Vis, a neural translation model for automatically generating visualizations from given datasets. We formulate visualization generation as a sequence-to-sequence translation problem in which data is mapped to visualization specifications in a declarative language. We'll discuss how we train a multilayered attention-based encoder-decoder model on a corpus of visualization specifications. Qualitative results show that the model learns the vocabulary and syntax for valid visualization specifications, appropriate transformations, and how to use common data-selection patterns occurring within data visualizations. We'll describe how Data2Vis generates high-quality visualizations comparable to manual efforts in a fraction of the time, and how it has the potential to learn more complex visualization strategies at scale. We will also provide guidance on training such a model using the Cloudera Datascience Workbench and explore uses for Data2Vis within visualization tools.  Back
 
Topics:
AI and DL Research, AI Application Deployment and Inference, Programming Languages
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9140
Streaming:
Download:
Share:
 
Abstract:
We'll discuss an implementation of GPU convolution that favors coalesced accesses without requiring prior data transformations. Convolutions are the core operation of deep learning applications based on convolutional neural networks. Current ...Read More
Abstract:

We'll discuss an implementation of GPU convolution that favors coalesced accesses without requiring prior data transformations. Convolutions are the core operation of deep learning applications based on convolutional neural networks. Current GPU architectures are typically used for training deep CNNs, but some state-of-the-art implementations are inefficient for some commonly used network configurations. We'll discuss experiments that used our new implementation, which yielded notable performance improvements including up to 2.29X speedups in a wide range of common CNN configurations.

  Back
 
Topics:
AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9218
Streaming:
Download:
Share:
 
Abstract:
Learning long-term dependencies in extended temporal sequences requires credit assignment to events far in the past. The most common method for training recurrent neural networks, backpropagation through time, requires credit information to be propag ...Read More
Abstract:
Learning long-term dependencies in extended temporal sequences requires credit assignment to events far in the past. The most common method for training recurrent neural networks, backpropagation through time, requires credit information to be propagated backwards through every single step of the forward computation, potentially over thousands or millions of time steps. We'll describe how this becomes computationally expensive or even infeasible when used with long sequences. Although biological brains are unlikely to perform such detailed reverse replay over very long sequences of internal states, humans often reminded of past memories or mental states associated with their current mental states. We'll discuss the hypothesis that such memory associations between past and present could be used for credit assignment through arbitrarily long sequences, propagating the credit assigned to the current state to the associated past state.  Back
 
Topics:
AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9251
Streaming:
Download:
Share:
 
Abstract:
Learn how to apply deep learning technologies for building robust and scalable dialogue systems with deeper understanding of the classic pipelines and final out more about the benchmark of models of prior work. We'll give an overview of dialogue res ...Read More
Abstract:
Learn how to apply deep learning technologies for building robust and scalable dialogue systems with deeper understanding of the classic pipelines and final out more about the benchmark of models of prior work. We'll give an overview of dialogue research and details state-of-the-art end-to-end neural dialogue systems for both task-oriented and social chit-chat conversations.  Back
 
Topics:
AI and DL Research, Speech and Language Processing
Type:
Tutorial
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9276
Streaming:
Download:
Share:
 
Abstract:
We'll discuss Alibaba's PAI tensor accelerator and optimizer (PAI-Tao), an elaborately implemented and optimized AI engine for deep learning training and inference tasks. PAI-Tao is designed with a data-driven and compiler-oriented approach. It per ...Read More
Abstract:
We'll discuss Alibaba's PAI tensor accelerator and optimizer (PAI-Tao), an elaborately implemented and optimized AI engine for deep learning training and inference tasks. PAI-Tao is designed with a data-driven and compiler-oriented approach. It periodically collects online running statistics to provide insights for optimization and uses collected statistics to help drive the real optimization work. We'll outline how PAI-Tao's compiler-oriented design can better accommodate diversified and fast-changing AI workloads.  Back
 
Topics:
AI and DL Research, Accelerated Data Science
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9280
Streaming:
Download:
Share:
 
Abstract:
Training and tuning models with lengthy training cycles like those in deep learning can be extremely expensive and may sometimes involve techniques that degrade performance. We'll explore recent research on optimization strategies to efficiently tun ...Read More
Abstract:
Training and tuning models with lengthy training cycles like those in deep learning can be extremely expensive and may sometimes involve techniques that degrade performance. We'll explore recent research on optimization strategies to efficiently tune these types of deep learning models. We will provide benchmarks and comparisons to other popular methods for optimizing the models, and we'll recommend valuable areas for further applied research.  Back
 
Topics:
AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9313
Streaming:
Download:
Share:
 
Abstract:
We will discuss a deep learning-based method for improving the quality of 3D reconstruction performed by time-of-flight cameras. Scene motion, multiple reflections, and sensor noise introduce artifacts in the depth reconstruction performed by these s ...Read More
Abstract:
We will discuss a deep learning-based method for improving the quality of 3D reconstruction performed by time-of-flight cameras. Scene motion, multiple reflections, and sensor noise introduce artifacts in the depth reconstruction performed by these sensors. We'll explain our proposed two-stage, deep-learning approach to address all of these sources of artifacts simultaneously. We'll also introduce FLAT, a synthetic dataset of 2000 ToF measurements that capture all of these nonidealities and can be used to simulate different hardware. Using the Kinect camera as a baseline, we show improved reconstruction errors on simulated and real data, as compared with state-of-the-art methods.  Back
 
Topics:
AI and DL Research, Computer Vision
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9318
Streaming:
Download:
Share:
 
Abstract:
In recent year, there is growing interest in advancing AI in the reasoning field, as reasoning is one of the main abilities associated with intelligence. Deep learning performs exceptionally well at pattern recognition and in recent times, it is adva ...Read More
Abstract:
In recent year, there is growing interest in advancing AI in the reasoning field, as reasoning is one of the main abilities associated with intelligence. Deep learning performs exceptionally well at pattern recognition and in recent times, it is advancing into the reasoning field, from relational networks for question answering and transparency-by-design networks for visual reasoning. In this talk, we will share an alternate and complementary paradigm for performing reasoning with a type theoretic approach.  Back
 
Topics:
AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9383
Streaming:
Download:
Share:
 
Abstract:
Learn how to achieve real-world speedup of neural networks using structural sparsity. Structural sparsity reduces the number of weights and computations in a way that's suitable for hardware acceleration. Over-parameterized neural networks waste mem ...Read More
Abstract:
Learn how to achieve real-world speedup of neural networks using structural sparsity. Structural sparsity reduces the number of weights and computations in a way that's suitable for hardware acceleration. Over-parameterized neural networks waste memory and energy. Techniques like pruning or factorization can alleviate this during inference but they often increase training time, and achieving real-world speedups remains difficult. We'll explain how biology-inspired techniques can reduce the number of weights from quadratic to linear in the number of neurons. Compared to fully connected neural networks, these structurally sparse neural networks achieve large speedups during both training and inference, while maintaining or even improving model accuracy. We'll discuss hardware considerations and results for feed-forward and recurrent networks.  Back
 
Topics:
AI and DL Research, Algorithms and Numerical Techniques
Type:
Tutorial
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9389
Streaming:
Download:
Share:
 
Abstract:
MATLAB's deep learning, visualization, and C++/CUDA code generation technology make it a uniquely complete solution for your entire AI workflow. In MATLAB, you can easily manage data, perform complex image and signal processing, prototype and train ...Read More
Abstract:
MATLAB's deep learning, visualization, and C++/CUDA code generation technology make it a uniquely complete solution for your entire AI workflow. In MATLAB, you can easily manage data, perform complex image and signal processing, prototype and train deep networks, and deploy to your desktop, embedded or cloud environments. Using GPU Coder technology MATLAB generates CUDA kernels that optimize loops and memory access, and C++ that leverages cuDNN and TensorRT, providing the fastest deep network inference of any framework. With MATLAB's NVIDIA docker container available through the NVIDIA GPU Cloud, you can now easily access all this AI power, deploy it in your cloud or DGX environment, and get up and running in seconds. In this presentation we will demonstrate a complete end-to-end workflow that starts from 'docker run', prototypes and trains a network on a multi-GPU machine in the cloud, and ends with a highly optimized inference engine to deploy to data centers, clouds, and embedded devices.  Back
 
Topics:
AI and DL Research, Data Center and Cloud Infrastructure
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9469
Streaming:
Download:
Share:
 
Abstract:
New to ML? Want to get started using TensorFlow together with GPUs? We will cover how you should use TensorFlow APIs to define and train your models, and discuss best practices for distributing the training workloads to multiple GPUs. We will also lo ...Read More
Abstract:
New to ML? Want to get started using TensorFlow together with GPUs? We will cover how you should use TensorFlow APIs to define and train your models, and discuss best practices for distributing the training workloads to multiple GPUs. We will also look at why GPUs are so great for machine learning workloads. This talk is appropriate for beginners who want to learn what TensorFlow can do.  Back
 
Topics:
AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9517
Streaming:
Share:
 
Abstract:
Learn about object detection and how GPUs are boosting the field in the PyTorch framework. As part of our talk, we'll discuss GPU implementation, including the efficiency-accuracy tradeoff and cluster deployment. We'll also delve into the latest ob ...Read More
Abstract:
Learn about object detection and how GPUs are boosting the field in the PyTorch framework. As part of our talk, we'll discuss GPU implementation, including the efficiency-accuracy tradeoff and cluster deployment. We'll also delve into the latest object-detection research.  Back
 
Topics:
AI and DL Research, Computer Vision
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9551
Streaming:
Download:
Share:
 
Abstract:
Learn how to quickly build robust deep neural networks for visual-recognition tasks using information generated directly from the human brain. We will present a novel active learning framework that combines fast image presentation, real-time brainwav ...Read More
Abstract:
Learn how to quickly build robust deep neural networks for visual-recognition tasks using information generated directly from the human brain. We will present a novel active learning framework that combines fast image presentation, real-time brainwave classification, and the use of classification score for optimizing the loss function. We'll share examples of using the proposed framework on GPUs to train neural networks and show that our solution provides faster convergence and higher performance than traditional methods.  Back
 
Topics:
AI and DL Research, Accelerated Data Science
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9554
Streaming:
Download:
Share:
 
Abstract:
We'll introduce a method for constructing an accurate prediction model from limited data in machine learning, one of the most important tasks in machine learning. We'll discuss unsupervised domain adaptation for open set data and a visual question- ...Read More
Abstract:
We'll introduce a method for constructing an accurate prediction model from limited data in machine learning, one of the most important tasks in machine learning. We'll discuss unsupervised domain adaptation for open set data and a visual question-generation method to acquire knowledge of unknown object categories.  Back
 
Topics:
AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9598
Streaming:
Download:
Share:
 
Abstract:
We'll describe joint NVIDIA-Amazon work to build AI assistive tools for the caretakers of the Amazon Biospheres. We train deep autoencoders on time-series sensor streams for monitoring anomalies in the micro-climate. Our talk will cover how we deplo ...Read More
Abstract:
We'll describe joint NVIDIA-Amazon work to build AI assistive tools for the caretakers of the Amazon Biospheres. We train deep autoencoders on time-series sensor streams for monitoring anomalies in the micro-climate. Our talk will cover how we deploy convolutional architectures for tracking plant stress levels using time-lapse vision models. We'll outline how we try to use best practices for edge-to-cloud AI and how we built a workflow to train models on EC2.P3 instances (NVIDIA Tesla V100 GPUs on AWS SageMaker). We'll also discuss how we optimize models for inference using TensorRT and subsequently deploy those models on NVIDIA Jetson TX2s for the biosphere.  Back
 
Topics:
AI and DL Research, Intelligent Machines and IoT, AI Application Deployment and Inference
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9627
Streaming:
Download:
Share:
 
Abstract:
We'll introduce several deep angular/hyperpherical learning frameworks and their applications in computer vision. Deep angular/hyperspherical learning provides state-of-the-art performance for general image classification and face recognition proble ...Read More
Abstract:
We'll introduce several deep angular/hyperpherical learning frameworks and their applications in computer vision. Deep angular/hyperspherical learning provides state-of-the-art performance for general image classification and face recognition problems. We'll talk about the motivation behind this type of learning and introduce some relevant variants under this framework. Deep hyperspherical learning has diverse applications in computer vision, and can also be used for learning neural network architectures and improving neural network generalization. We'll also discuss a few open problems in this framework and talk about some potential applications.  Back
 
Topics:
AI and DL Research, Computer Vision
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9631
Streaming:
Download:
Share:
 
Abstract:
Learn why deep learning scales so well and how to apply it to important open problems. Deep learning has enabled rapid progress in diverse problems in vision, speech, and beyond. Driving this progress are breakthroughs in algorithms that can harness ...Read More
Abstract:
Learn why deep learning scales so well and how to apply it to important open problems. Deep learning has enabled rapid progress in diverse problems in vision, speech, and beyond. Driving this progress are breakthroughs in algorithms that can harness massive datasets and powerful compute accelerators like GPUs. We'll combine theoretical and experiment insights to help explain why deep learning scales predictably with bigger datasets and faster computers. We'll also show how some problems are relatively easier than others and how to tell the difference. Learn about examples of open problems that cannot be solved by individual computers, but are within reach of the largest machines in the world. We'll also make the case for optimizing data centers to run AI workloads. Finally, we'll outline a high-level architecture for an AI datacenter, and leave you with powerful tools to reach beyond human accuracy to confront some of the hardest open problems in computing.  Back
 
Topics:
AI and DL Research, HPC and AI
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9643
Streaming:
Share:
 
Abstract:
Academic design of deep neural networks has historically focused on maximizing accuracy at any cost. However, many practical applications have real-world constraints such as model size, computational complexity (FLOPs), or inference latency, as well ...Read More
Abstract:
Academic design of deep neural networks has historically focused on maximizing accuracy at any cost. However, many practical applications have real-world constraints such as model size, computational complexity (FLOPs), or inference latency, as well as physical hardware performance, that need to be considered. We'll discuss our MorphNet solution, an approach to automate the design of neural nets with constraint-specific and hardware-specific tradeoffs while being lightweight and scalable to large data sets. We show how MorphNet can be used to design neural nets that reduce model size, FLOP count, or inference latency with the same accuracy across different domains such as ImageNet, OCR, and AudioSet. Finally, we show how MorphNet designs different architectures when optimizing for P100 and V100 platforms.  Back
 
Topics:
AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9645
Streaming:
Download:
Share:
 
Abstract:
In this talk, I'll discuss several semi-supervised learning applications from our recent work in applied deep learning research at NVIDIA. I'll first discuss video translation, which renders new scenes using models learned from real-world videos. W ...Read More
Abstract:
In this talk, I'll discuss several semi-supervised learning applications from our recent work in applied deep learning research at NVIDIA. I'll first discuss video translation, which renders new scenes using models learned from real-world videos. We take real world videos, analyze them using existing computer vision techniques such as pose estimation or semantic segmentation, and then train generative models to invert these poses or segmentations back to videos. In deployment, we then render novel sketches using these models. I'll then discuss work on large-scale language modeling, where a model trained to predict text, piece by piece, on a large dataset is then finetuned with small amounts of labeled data to solve problems like emotion classification. Finally, I'll discuss WaveGlow, our flow-based generative model for the vocoder stage of speech synthesis, that combines a simple log-likelihood based training procedure with very fast and efficient inference. Because semi-supervised learning allows us to try tackling problems where large amounts of labels would be prohibitively expensive to create, it opens the scope of problems to which we can apply machine learning.  Back
 
Topics:
AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9686
Streaming:
Download:
Share:
 
Abstract:
We'll talk about how we're incorporating physics into deep learning algorithms. Standard deep learning algorithms are based on a function-fitting approach that does not exploit any domain knowledge or constraints. This makes them unsuitable for app ...Read More
Abstract:
We'll talk about how we're incorporating physics into deep learning algorithms. Standard deep learning algorithms are based on a function-fitting approach that does not exploit any domain knowledge or constraints. This makes them unsuitable for applications like robotics that require safety or stability guarantees. These algorithms also require large amounts of labeled data, which is not readily available. We'll discuss how we're overcoming these limitations by infusing physics into deep learning algorithms, and how we're applying this to stable landing of quadrotor drones. We've developed a robust deep learning-based nonlinear controller called Neural-Lander, which learns ground-effect aerodynamic forces that are hard to model. We'll also touch on how Neural-Lander can land significantly faster while maintaining stability.  Back
 
Topics:
AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9732
Streaming:
Download:
Share:
 
Abstract:
Learn about tensors, higher-order extensions of matrices that can incorporate multiple modalities and encode higher-order relationships in data. After an introduction to tensor methods, we will discuss which tensor methods can be used in deep learnin ...Read More
Abstract:
Learn about tensors, higher-order extensions of matrices that can incorporate multiple modalities and encode higher-order relationships in data. After an introduction to tensor methods, we will discuss which tensor methods can be used in deep learning and in probabilistic modeling. We'll show how tensor contractions, which are extensions of matrix products, provide high rates of compression in a variety of neural network models. We'll also demonstrate the use of tensors for document categorization at scale through probabilistic topic models. These are available in a python library called Tensorly that provides a high-level API for tensor methods and deep tensorized architectures.  Back
 
Topics:
AI and DL Research
Type:
Tutorial
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9733
Streaming:
Download:
Share:
 
Abstract:
We'll present our study on GPU optimization for deep learning with limited computational resources and share our tips and tricks for building a state-of-the-art Visual Question Answering (VQA) system. Learn about technical implementations of deep le ...Read More
Abstract:
We'll present our study on GPU optimization for deep learning with limited computational resources and share our tips and tricks for building a state-of-the-art Visual Question Answering (VQA) system. Learn about technical implementations of deep learning algorithms with GPU hardware utilization, including delayed updates and mixed-precision training, to deal with limited hardware resources while reduce training time and memory usage. We'll describe our experience designing a winning architecture for the VQA Challenge 2018 by applying deep learning tactics such as multi-level multi-modal fusion, parameter-interaction learning, and end-to-end optimization. Our techniques are all heavy computing tasks, so GPU programming plays an important role in advancing our work. We'll also provide convincing empirical proofs and a practical demonstration of a VQA application.  Back
 
Topics:
AI and DL Research, AI Application Deployment and Inference
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9824
Streaming:
Download:
Share:
 
Abstract:
We'll discuss learning to synthesize object instances such as a person or car in both 2D and 3D scenes. We will introduce our work we presented at NeurIPS 2018 on context-aware synthesis and placement of object instances. We propose a generative mod ...Read More
Abstract:
We'll discuss learning to synthesize object instances such as a person or car in both 2D and 3D scenes. We will introduce our work we presented at NeurIPS 2018 on context-aware synthesis and placement of object instances. We propose a generative model that learns to generate and insert an object instance into an image in a semantically coherent manner. In particular, we represent object instances using masks and learn to insert them into semantic label maps of images. Our talk will also cover our recent work around putting humans in a scene and learning affordance in 3D indoor environments. This extends the learning of context from 2D to 3D scenes in which the synthesized objects are semantically coherent and geometrically correct. We'll show that both projects add technical insights and have potential applications in content creation.  Back
 
Topics:
AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9959
Streaming:
Download:
Share:
 
Abstract:
Join a special presentation from our 2018-2019 Graduate Fellowship recipients to learn what's next from the world of research and academia. Sponsored projects involve a variety of technical challenges, including topics such as 3D scene understanding ...Read More
Abstract:
Join a special presentation from our 2018-2019 Graduate Fellowship recipients to learn what's next from the world of research and academia. Sponsored projects involve a variety of technical challenges, including topics such as 3D scene understanding, new programming models for tensor computations, HPC physics simulations for astrophysics, deep learning algorithms for AI natural language learning, and cancer diagnosis. We believe that theses students will lead the future in our industry and we're proud to support the 2018-2019 NVIDIA Graduate Fellows. For more information on the NVIDIA Graduate Fellowship program, visit www.nvidia.com/en-us/research/graduate-fellowships.  Back
 
Topics:
AI and DL Research, Virtual Reality and Augmented Reality, Graphics and AI, Computational Biology and Chemistry, Computer Vision
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9976
Streaming:
Download:
Share:
AI for Gaming
Presentation
Media
Abstract:
We will share the nitty-gritty details of DICE's Battlefield V ray-traced reflection implementation. Well start with a brief primer on DXR, so previous experience with the API is not required to understand this talk. Our talk will cover ever ...Read More
Abstract:

We will share the nitty-gritty details of DICE's Battlefield V ray-traced reflection implementation. Well start with a brief primer on DXR, so previous experience with the API is not required to understand this talk. Our talk will cover everything from shader generation to denoising that is necessary to ship a high-quality title with ray-traced reflections. Well show plenty of eye-candy to demonstrate how if you do everything right and are willing to invest a lot of blood, sweat and tears, how our implementation just works. 

  Back
 
Topics:
AI for Gaming
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S91023
Streaming:
Download:
Share:
 
Abstract:
We will talk about Q2VKPT, the Vulkan-based renderer for Quake 2 that uses hardware accelerated path tracing and advanced spatiotemporal denoising. We will cover some of the implementation details, including things like importance sampling, ...Read More
Abstract:

We will talk about Q2VKPT, the Vulkan-based renderer for Quake 2 that uses hardware accelerated path tracing and advanced spatiotemporal denoising. We will cover some of the implementation details, including things like importance sampling, lighting, materials, and the denoising filters. In addition, we will discuss the challenges of using a physically based renderer with the assets from a game released over two decades ago.

  Back
 
Topics:
AI for Gaming
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S91046
Streaming:
Download:
Share:
 
Abstract:
Learn about a building block to render ray-traced global illumination in real-time games. Restricting path tracing to a small number of paths per pixel for performance reasons rarely achieves a satisfactory image quality for scenes of interest. Howev ...Read More
Abstract:
Learn about a building block to render ray-traced global illumination in real-time games. Restricting path tracing to a small number of paths per pixel for performance reasons rarely achieves a satisfactory image quality for scenes of interest. However, path space filtering may dramatically improve the visual quality by sharing information across vertices of paths classified as nearby. Although these contributions can be filtered in path space and beyond the first intersection, searching nearby paths is more expensive than filtering in screen space. We'll explain how we overcame this performance penalty by storing and looking up the required information in a hash map, using hash keys constructed from jittered and quantized information, such that only a single query may replace costly neighborhood searches.  Back
 
Topics:
AI for Gaming, Rendering and Ray Tracing, Real-Time Graphics
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9462
Streaming:
Download:
Share:
 
Abstract:
We'll discuss Substance Alchemist, a tool that will allow users to manage material collections and create new materials from pictures, scans, or pre-existing materials. We will detail the different facets of Alchemist and explain how GPU-Accelerated ...Read More
Abstract:
We'll discuss Substance Alchemist, a tool that will allow users to manage material collections and create new materials from pictures, scans, or pre-existing materials. We will detail the different facets of Alchemist and explain how GPU-Accelerated AI will enhance material creation, with a deep dive into the delighter features that leverage TensorCore. The tool was first demonstrated at SIGGRAPH 2018.  Back
 
Topics:
AI for Gaming, Animation and VFX, Digital Product Design & Styling
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9739
Streaming:
Download:
Share:
 
Abstract:
Find out what it takes to bring the benefits of AI to applications that run at interactive speeds. We'll explain how to think about inference performance throughout the AI development pipeline, covering topics like network design, training considera ...Read More
Abstract:
Find out what it takes to bring the benefits of AI to applications that run at interactive speeds. We'll explain how to think about inference performance throughout the AI development pipeline, covering topics like network design, training considerations, debugging, profiling performance, and optimizing GPU-Based code for fastest throughput using Tensor Core acceleration. This talk is aimed at application programmers and other AI practitioners, but it should be accessible to program designers and project managers interested in incorporating AI features into their products.  Back
 
Topics:
AI for Gaming, AI Application Deployment and Inference, Real-Time Graphics
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9765
Streaming:
Download:
Share:
 
Abstract:
We'll provide a deep dive into implementing irradiance fields with moment visibility for local and cloud rendering using RTX. This is a real-time, fully dynamic diffuse global illumination solution that prevents light leaks and screen-space noise. T ...Read More
Abstract:
We'll provide a deep dive into implementing irradiance fields with moment visibility for local and cloud rendering using RTX. This is a real-time, fully dynamic diffuse global illumination solution that prevents light leaks and screen-space noise. The technology can operate on local GI on a Turing GPU for PC games or a streaming cloud GI from a Turing-powered server for VR and mobile gaming.  Back
 
Topics:
AI for Gaming, Data Center and Cloud Infrastructure, Rendering and Ray Tracing
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9900
Streaming:
Download:
Share:
 
Abstract:
We'll present Isaac Gym, a platform for distributed high-performance reinforcement learning training, research in robotics, animation, and biomechanics. Isaac Gym supports different rendering and simulation, including Flex and PhysX backends. We'll ...Read More
Abstract:
We'll present Isaac Gym, a platform for distributed high-performance reinforcement learning training, research in robotics, animation, and biomechanics. Isaac Gym supports different rendering and simulation, including Flex and PhysX backends. We'll discuss how GPU-Accelerated high fidelity physics simulation can simulate not only rigid but also deformable soft-bodies, cloth, ropes and liquids, and interaction between these elements. Our reinforcement learning training pipeline is also GPU-Accelerated and we provide fast parallel multi-camera rendering support for tasks involving vision. We'll explain how a Python wrapper simplifies and speed prototyping, and allows us to perform many experiments and set benchmarks. We'll show some of the most impressive training results in classical and new challenging robotics and locomotion training environments in Isaac Gym and share information about the performance and scalability of our GPU-Accelerated simulation and training pipeline.  Back
 
Topics:
AI for Gaming, AI and DL Research, Intelligent Machines and IoT
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9918
Streaming:
Download:
Share:
 
Abstract:
We will examine tools and technologies that NVIDIA's GameWorks team is building to leverage the power of deep learning for content creation, and we'll demonstrate how we're combining some of these techniques with traditional character animation fo ...Read More
Abstract:
We will examine tools and technologies that NVIDIA's GameWorks team is building to leverage the power of deep learning for content creation, and we'll demonstrate how we're combining some of these techniques with traditional character animation for applications in our Isaac robotics simulator. We will also talk about how to use GPUs for high-performance runtime inferencing of these networks for games or real-time VFX scenarios. We'll also eview some of the latest research in the field, particularly that involving physics-based simulation and reinforcement learning, and demonstrate some of NVIDIA's latest work designed to simplify the process of training RL agents for these tasks.  Back
 
Topics:
AI for Gaming
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9963
Streaming:
Download:
Share:
 
Abstract:
This session will cover how Nixxes and NVIDIA added ray-traced shadows for directional, spot, point, and area light sources. We'll cover what went well and what didn't in all aspects of the work, including BVH construction and implicatio ...Read More
Abstract:

This session will cover how Nixxes and NVIDIA added ray-traced shadows for directional, spot, point, and area light sources. We'll cover what went well and what didn't in all aspects of the work, including BVH construction and implications on content, as well as tracing and denoising the results.

  Back
 
Topics:
AI for Gaming
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9984
Streaming:
Download:
Share:
 
Abstract:
We'll discuss ray-tracing in Metro Exodus, covering topics like stochastic effects are your friends, and how ray-tracing could improve the non-RT rendering pipeline in Metro and DXR integration without massive changes in the engine. We'll touch on ...Read More
Abstract:
We'll discuss ray-tracing in Metro Exodus, covering topics like stochastic effects are your friends, and how ray-tracing could improve the non-RT rendering pipeline in Metro and DXR integration without massive changes in the engine. We'll touch on deferred lighting for hit positions, explore global illumination and the possibility of less-than-one-ray-per-pixel computer graphics and compare more rays versus more denoising. Our talk will also explore other ray-tracking effects and tips and tricks for denoising at full resolution and full performance.  Back
 
Topics:
AI for Gaming
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9985
Streaming:
Download:
Share:
 
Abstract:
We will discuss the latest features of PhysX 4, NVIDIA's latest open-sourced PhysX version. We'll explain new techniques and how to use them, and we'll provide details about the performance and accuracy trade-offs of these techniques and their app ...Read More
Abstract:
We will discuss the latest features of PhysX 4, NVIDIA's latest open-sourced PhysX version. We'll explain new techniques and how to use them, and we'll provide details about the performance and accuracy trade-offs of these techniques and their applicability to game developers. Our talk will provide examples of how these techniques can be used to improve simulation quality and performance in a wide range of gaming applications. We'll also discuss the feature set and simulation integrity of the PhysX Vehicles SDK.  Back
 
Topics:
AI for Gaming
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9990
Streaming:
Download:
Share:
AI in Healthcare
Presentation
Media
Abstract:
We will talk about how we're using AI to solve problems in healthcare, and specifically in cardiovascular imaging. We built foundational tools for view classification in echocardiography, using NVIDIA GPUs to provide classification- and segmentation ...Read More
Abstract:
We will talk about how we're using AI to solve problems in healthcare, and specifically in cardiovascular imaging. We built foundational tools for view classification in echocardiography, using NVIDIA GPUs to provide classification- and segmentation-based diagnosis of cardiovascular disease. We'll show how we apply this work toward certain unmet needs in cardiology, using methods that are highly applicable across several fields in medicine.  Back
 
Topics:
AI in Healthcare, Medical Imaging and Radiology
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9104
Streaming:
Share:
 
Abstract:
Learn about the key types of clinical use cases for AI methods in medical imaging beyond image classification that will ultimately improve medical practice. We'll explain the critical challenges and progress in applying AI in these applications, and ...Read More
Abstract:
Learn about the key types of clinical use cases for AI methods in medical imaging beyond image classification that will ultimately improve medical practice. We'll explain the critical challenges and progress in applying AI in these applications, and describe the types of medical imaging and the clinical applications for deep learning to improve image interpretation. We will also talk about recent AI projects that tackle the challenging problem of clinical prediction with innovative approaches that provide explanations about AI model predictions to improve clinician acceptance.  Back
 
Topics:
AI in Healthcare, AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9115
Streaming:
Download:
Share:
 
Abstract:
We'll demonstrate how different kinds of AI techniques can be used across all four stages of drug development preclinical, clinical, regulatory, and commercial. Each stage requires a huge volume of mostly unstructured data that must be analyzed to g ...Read More
Abstract:
We'll demonstrate how different kinds of AI techniques can be used across all four stages of drug development preclinical, clinical, regulatory, and commercial. Each stage requires a huge volume of mostly unstructured data that must be analyzed to generate insights. We'll provide several examples of our methods, including using GPUs to accelerate segmentation and image processing in information extraction from PDFs, HTML files, and images combining CNN and LSTM. Other examples include using GPUs to accelerate training to identify biomedical entities and resolve them to correct type with a combination of CRF and RNN. We're also using GPUs to accelerate graph traversals in mapping connections among millions of biomedical concepts in real time.  Back
 
Topics:
AI in Healthcare, AI and DL Business Track (high level)
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9614
Streaming:
Download:
Share:
 
Abstract:
We'll describe our experience using Unified Memory for full-resolution medical image AI applications. Although researchers are quickly embracing deep learning-based methods for medical image analysis, constraints on computer hardware and systems sof ...Read More
Abstract:
We'll describe our experience using Unified Memory for full-resolution medical image AI applications. Although researchers are quickly embracing deep learning-based methods for medical image analysis, constraints on computer hardware and systems software have begun to show. One issue is GPU memory. Although GPUs dramatically accelerate deep neural network training, they have less memory compared with CPUs. That limits the choice of input image size, neural network architecture, and batch size for deep learning-based methods, which leads to inferior results. We'll show how NVIDIA CUDA is an ideal solution because its Unified Memory architecture allows GPUs to access system memory. We'll discuss how lifting size limits for models, batch size and input size affects training throughput and model performance.  Back
 
Topics:
AI in Healthcare
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9668
Streaming:
Download:
Share:
 
Abstract:
We'll talk about using GPU computing to improve the diagnosis and treatment of neurological disorders. After identifying actionable neurological disorders, we'll recast them as tractable deep learning problems and discuss techniques to solve them. ...Read More
Abstract:
We'll talk about using GPU computing to improve the diagnosis and treatment of neurological disorders. After identifying actionable neurological disorders, we'll recast them as tractable deep learning problems and discuss techniques to solve them. We'll discuss the architecture of an on-prem deep learning cluster in a hospital work environment, and describe the creation of production-ready systems for clinical deployment. Our talk will focus on computer vision techniques, radiological computer-assisted diagnosis, and clinical decision support in the intensive care unit. We will also discuss the question of flat minima and generalization in a medical context, and cover potential solutions and model training.  Back
 
Topics:
AI in Healthcare, Medical Imaging and Radiology
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9899
Streaming:
Download:
Share:
 
Abstract:
We'll talk about how AI can reinvent the workflow for radiology. Understanding the current workflow and challenges in radiology is important when developing potential AI solutions, but developers must avoid simply replacing existing workflow steps w ...Read More
Abstract:
We'll talk about how AI can reinvent the workflow for radiology. Understanding the current workflow and challenges in radiology is important when developing potential AI solutions, but developers must avoid simply replacing existing workflow steps with AI. We'll discuss how using AI to refashion the workflow could help radiologists and physicians in other diagnostic imaging specialties deliver more effective, personalized, cost-effective, and accessible care to patients. We'll describe challenges specific to patient care in radiology, brainstorm solutions, and describe AI initiatives we're piloting.  Back
 
Topics:
AI in Healthcare, AI Application Deployment and Inference
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9924
Streaming:
Download:
Share:
 
Abstract:
We'll talk about overcoming the barrier to obtaining adequate data to generate highly accurate models in medical imaging. Deep learning has proven valuable in creating analytic tools for medical imaging that have been cleared in multiple countries a ...Read More
Abstract:
We'll talk about overcoming the barrier to obtaining adequate data to generate highly accurate models in medical imaging. Deep learning has proven valuable in creating analytic tools for medical imaging that have been cleared in multiple countries and are used in daily clinical work. But one barrier to additional solutions has been a lack of available data. Data are often difficult to obtain, especially in the quantities needed to generate highly accurate models. We'll discuss data augmentation techniques and explain how three different institutions have overcome this barrier.  Back
 
Topics:
AI in Healthcare, Advanced AI Learning Techniques, Medical Imaging and Radiology
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9995
Streaming:
Download:
Share:
Accelerated Data Science
Presentation
Media
Abstract:
At Uber, real-time analytics plays a critical role in gaining business insights and operational efficiency. It also enables us to make real-time data-driven decisions to improve user experiences on our transportation platform. GPU technology has adva ...Read More
Abstract:
At Uber, real-time analytics plays a critical role in gaining business insights and operational efficiency. It also enables us to make real-time data-driven decisions to improve user experiences on our transportation platform. GPU technology has advanced significantly over the years, making it a perfect fit for large-scale computation and data processing in parallel. We'll describe why we leverage GPUs to solve the real-time analytics problem at Uber and how we designed our system to fully utilize the GPU's parallelization power and minimize its drawbacks. We will also share learnings in GPU programming during the development of AresDB.  Back
 
Topics:
Accelerated Data Science, Performance Optimization
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S91021
Streaming:
Download:
Share:
 
Abstract:
The core of RAPIDS is CUDA DataFrame (cuDF), a library that provides Pandas-like DataFrame (a columnar data structure) functionality with GPU acceleration. cuDF provides a Python interface for use in existing data science workflows, and undernea ...Read More
Abstract:

The core of RAPIDS is CUDA DataFrame (cuDF), a library that provides Pandas-like DataFrame (a columnar data structure) functionality with GPU acceleration. cuDF provides a Python interface for use in existing data science workflows, and underneath cuDF is libcuDF, an open-source CUDA C++ library that provides a column data structure and algorithms to operate on these columns, such as filtering, selection, sorting, joining, and groupby. In this talk you will learn about some of the C++ and CUDA internals of libcuDF. This talk will cover how we perform run-time type dispatch on type-erased data structures to enable operating on a variety of data types and interface with dynamic languages like Python. Well describe how and why we built a pool allocator for CUDA device memory to massively improve performance on multi-GPU systems. And well dive into GPU algorithms we use for multi-column database operations like groupby and join. If you are interested in using GPU DataFrames via libcuDFs C/C++ interface, or if you are interested in contributing to the cuDF / libcuDF open source project, then this talk is for you.

  Back
 
Topics:
Accelerated Data Science, RAPIDS
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S91043
Streaming:
Download:
Share:
 
Abstract:
We'll discuss the GPU Open Analytics Initiative, an effort to develop a GPU data frame that can handle a large-scale data-analytics workflow and support out-of-core cases in which the data is larger than GPU memory. We'll describe how we divided th ...Read More
Abstract:
We'll discuss the GPU Open Analytics Initiative, an effort to develop a GPU data frame that can handle a large-scale data-analytics workflow and support out-of-core cases in which the data is larger than GPU memory. We'll describe how we divided the problem into two parts, developing an elementary single-GPU data frame to handle in-memory use cases, and then combining multiple single-GPU data frames into a distributed multi-GPU data frame for out-of-core use cases. We'll briefly introduce our distributed GPU data frame and its capabilities. We'll then explain how we scaled out by using Dask, a distributed computation framework in Python, to orchestrate the single-GPU data frames and achieve out-of-core capability with minimal effort. Our idea can be generalized to build custom distributed GPU computation by composing single-GPU libraries.  Back
 
Topics:
Accelerated Data Science, Tools and Libraries
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9449
Streaming:
Download:
Share:
 
Abstract:
Connected vehicles generate a treasure trove of anonymized data with unlimited potential for the automobile industry. Vehicle usage, fleet movement, mobility patterns, telematics and spatiotemporal data are invaluable for everyone from development to ...Read More
Abstract:
Connected vehicles generate a treasure trove of anonymized data with unlimited potential for the automobile industry. Vehicle usage, fleet movement, mobility patterns, telematics and spatiotemporal data are invaluable for everyone from development to manufacturing. Hear how the BMW Group has built dashboards with OmniSci Immerse to visualize this telematics data. See a live demonstration of this application as it's used to visualize spatiotemporal telematics data and interact with massive datasets-in-motion with near-zero latency. Join Caroline Persson, Data Scientist at the BMW Group, and Todd Mostak, CEO of OmniSci, as they explain how the parallel processing power of GPUs unlocks a wealth of use cases across manufacturing and other major industries, driving operational analytics, geospatial analytics, and data science. Persson will provide an example use case and show how automobile manufacturers can ingest huge volumes of streaming telematics data coming from vehicles and apply machine learning models to descriptive driver behavior and patterns.  Back
 
Topics:
Accelerated Data Science, Deep Learning and AI Frameworks
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9537
Streaming:
Download:
Share:
 
Abstract:
Most large companies use online analytical processing (OLAP) to gain insight from available data and guide business decisions. To support time-critical business decisions, companies must answer queries as quickly as possible. For OLAP, the performanc ...Read More
Abstract:
Most large companies use online analytical processing (OLAP) to gain insight from available data and guide business decisions. To support time-critical business decisions, companies must answer queries as quickly as possible. For OLAP, the performance bottlenecks are joins of large relations. GPUs can significantly accelerate these joins, but often the speed or memory capacity of a single GPU is not sufficient to join input tables or unable to do it quickly enough. We'll discuss how we're addressing these problems by proposing join algorithms that scale to multiple GPUs.  Back
 
Topics:
Accelerated Data Science, Performance Optimization
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9557
Streaming:
Download:
Share:
 
Abstract:
See how RAPIDS and the open source ecosystem are advancing data science. In this session, we will explore RAPIDS, the NEW open source data science platform from NVIDIA. Come learn how to get started leveraging these open-source libraries for fas ...Read More
Abstract:

See how RAPIDS and the open source ecosystem are advancing data science. In this session, we will explore RAPIDS, the NEW open source data science platform from NVIDIA. Come learn how to get started leveraging these open-source libraries for faster performance and easier development on GPUs. See the latest engineering work and new release features, including, benchmarks, roadmaps, and demos. Finally, hear how customers are leveraging RAPIDS in production, benefiting from early adoption, and outperforming CPU equivalents.

  Back
 
Topics:
Accelerated Data Science, RAPIDS
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9577
Streaming:
Download:
Share:
 
Abstract:
Graphs are a ubiquitous part of technology we use daily in systems like GPS graphs help find the shortest path between two points and in social networks, which use them to help users find friends. We'll explain why analyzing these vast netwo ...Read More
Abstract:

Graphs are a ubiquitous part of technology we use daily in systems like GPS graphs help find the shortest path between two points and in social networks, which use them to help users find friends. We'll explain why analyzing these vast networks with possibly billions of entries requires the computing power of GPUs. We'll then discuss the performance of graph algorithms on the GPU and show benchmarking results from several graph frameworks. We'll also cover the RAPIDS roadmap that will help unify these frameworks and make them easy to use and simple to deploy.

  Back
 
Topics:
Accelerated Data Science, Algorithms and Numerical Techniques, RAPIDS
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9783
Streaming:
Download:
Share:
 
Abstract:
Some of the problems especially autonomous vehicles NVIDIA is tackling operate under heavy constraints: scalability, regulations or need for automation. In this presentation, we will show how we approach the problem so we can increase velocity and im ...Read More
Abstract:
Some of the problems especially autonomous vehicles NVIDIA is tackling operate under heavy constraints: scalability, regulations or need for automation. In this presentation, we will show how we approach the problem so we can increase velocity and improve the long-term outcome of our applications: how we enable automation, our strategy for leveraging large amounts of compute for diverse use cases, and also how we think about traceability, reproducibility and safety, with an eye on the future  Back
 
Topics:
Accelerated Data Science, Deep Learning and AI Frameworks
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9787
Streaming:
Download:
Share:
 
Abstract:
As the number of GPU-accelerated applications has multiplied, the need for better development tools and services have increased as well. Chief among such services is continuous integration (CI), which dramatically improves and speeds up the developme ...Read More
Abstract:
As the number of GPU-accelerated applications has multiplied, the need for better development tools and services have increased as well. Chief among such services is continuous integration (CI), which dramatically improves and speeds up the development lifecycle through automated builds and integration testing. CI for GPU-accelerated applications comes with its own set of challenges, but the rewards can be enormous. Join NVIDIA 's team as they walk through how they implemented CI by leaning on open source technologies such as Conda, Docker, and Jenkins, the lessons they learned in the process, and how other such systems should be built in the future.  Back
 
Topics:
Accelerated Data Science
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9788
Streaming:
Download:
Share:
 
Abstract:
Location intelligence is key to understanding areas such as property insights, environmental monitoring, disaster management and prevention, traffic flows, and customer behavior. We'll discuss our work involving Europe's property insuran ...Read More
Abstract:

Location intelligence is key to understanding areas such as property insights, environmental monitoring, disaster management and prevention, traffic flows, and customer behavior. We'll discuss our work involving Europe's property insurance sector, which has been disrupted by the growing use of comparison websites that require real-time quotations. To build deep learning models, large volumes of data from satellite images, 3D sensors, GPS-enabled devices, social media, and other sources must be merged using computationally intensive coordinate conversion and matching. We'll outline our solution, which uses 3D CNNs to estimate risk factors from color 3D virtual models of individual properties. We'll describe how we used RAPIDS and cover our entire process, from processing raw data, merging sources, generating and labeling colorized voxel cubes for training, to model building, inference, and final application.

  Back
 
Topics:
Accelerated Data Science, AI Application Deployment and Inference, Advanced AI Learning Techniques, RAPIDS
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9791
Streaming:
Download:
Share:
 
Abstract:
Radio frequency (RF) systems have become increasingly complex, and the number of connected devices is expected to increase. We'll discuss how deep learning within RF shows promise for dealing with a congested spectrum by enhancing reliability and si ...Read More
Abstract:
Radio frequency (RF) systems have become increasingly complex, and the number of connected devices is expected to increase. We'll discuss how deep learning within RF shows promise for dealing with a congested spectrum by enhancing reliability and simplifying the task of building effective wireless systems. Deep learning algorithms within RF technology show superior results, classifying signals well below the noise floor when compared to traditional signal processing methods. We'll describe how we've worked with partners to design a software-configurable wide-band RF transceiver system that can perform real-time DSP and deep learning with an NVIDIA GPU. We'll discuss RF system performance, RF training data collection, and software used to create applications. Additionally, we will present data demonstrating applications in deep learning enabled-RF technology.  Back
 
Topics:
Accelerated Data Science, Telecommunications, AI Application Deployment and Inference
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9792
Streaming:
Download:
Share:
 
Abstract:
Modern data science demands interactive exploration and analysis of large volumes of data. Learn how NVIDIA and RAPIDS take advantage of GPU acceleration by using libraries such as cuDF, cuIO, and cuString. The computational limits of CPUs are b ...Read More
Abstract:

Modern data science demands interactive exploration and analysis of large volumes of data. Learn how NVIDIA and RAPIDS take advantage of GPU acceleration by using libraries such as cuDF, cuIO, and cuString. The computational limits of CPUs are being realized. We'll how RAPIDS uses GPUs to accelerate existing workflows and enable workflows that were previously impossible. We'll cover cuDF's high-level architecture and its GPU use, and do a technical dive into cuDF internals such as the cuIO and cuString libraries. We'll also share testing and benchmarking results and reveal some of the new features and optimizations we're investigating for the future of RAPIDS and cuDF.

  Back
 
Topics:
Accelerated Data Science, Tools and Libraries, RAPIDS
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9793
Streaming:
Download:
Share:
 
Abstract:
Learn how RAPIDS uses Dask to scale to distributed clusters of machines. Dask, a library for scalable computing in Python, is known for scaling out popular PyData libraries like Numpy, Pandas, and Scikit-Learn. The GPU-Accelerated data science s ...Read More
Abstract:

Learn how RAPIDS uses Dask to scale to distributed clusters of machines. Dask, a library for scalable computing in Python, is known for scaling out popular PyData libraries like Numpy, Pandas, and Scikit-Learn. The GPU-Accelerated data science software stack RAPIDS also uses Dask to easily scale to multiple GPUs on a single node, and multiple nodes within a cluster. We'll explain how RAPIDS used Dask to scale out, discuss the challenges of integrating GPUs into the existing PyData stack, and describe how this work creates opportunities for Python users.

  Back
 
Topics:
Accelerated Data Science, RAPIDS, RAPIDS
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9797
Streaming:
Download:
Share:
 
Abstract:
Learn about BlazingSQL, our new, free GPU SQL engine built on RAPIDS open-source software. We will show multiple demo workflows using BlazingSQL to connect data lakes to RAPIDS tools. We'll explain how we dramatically accelerated our engine ...Read More
Abstract:

Learn about BlazingSQL, our new, free GPU SQL engine built on RAPIDS open-source software. We will show multiple demo workflows using BlazingSQL to connect data lakes to RAPIDS tools. We'll explain how we dramatically accelerated our engine and made it substantially more lightweight by integrating Apache Arrow into GPU memory and cuDF into RAPIDS. That made it easy to install and deploy BlazingSQL + RAPIDS in a matter of minutes. More importantly, we built a robust framework to help users bring data from data lakes into GPU-Accelerated workloads without having to ETL on CPU memory or separate GPU clusters. We'll discuss how that makes it possible to keep everything in the GPU while BlazingSQL manages the SQL ETL. RAPIDS can then take these results to continue machine learning, deep learning, and visualization workloads.

  Back
 
Topics:
Accelerated Data Science, RAPIDS
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9798
Streaming:
Download:
Share:
 
Abstract:
Walmart Labs has been charged with building the next-generation stores forecasting system, replacing the current system from JDA.   Due to the size of the forecasting problem 52 weekly forecasts for roughly 500 million store-item combi ...Read More
Abstract:

Walmart Labs has been charged with building the next-generation stores forecasting system, replacing the current system from JDA.   Due to the size of the forecasting problem 52 weekly forecasts for roughly 500 million store-item combinations, generated every week we realized early on that we would have to use GPU computing if we wanted to move beyond simple forecasting approaches such as exponential smoothing.   We have taken a multi-pronged approach to the problem of improving forecast accuracy while remaining within execution time windows using NVIDIA-supplied software such as XGBoost for forecasting, developing custom algorithms (some in CUDA) for various forecasting and forecasting-related processes, and moving to a RAPIDS-based feature generation pipeline.  At the moment, roughly 20% of our items are being forecasted by the new system, and we expect to have 100% item coverage by the end of the year. In this talk we will outline our forecasting strategy both from an algorithmic and from a computational perspective.  We will show how GPU computing has enabled us to significantly improve forecast accuracy, and highlight the key bottlenecks that we have been able to overcome.   We will provide runtime comparisons of CPU vs GPU-based algorithms on our real-world problems, and describe how GPU-based development works for us (hint: its easy to do.)   We will also describe our collaboration with NVIDIA, who have been extremely helpful, continuously refining their algorithms and tools to better meet the needs of industry, and what tools and capabilities we see being especially useful for our path forward. 

  Back
 
Topics:
Accelerated Data Science, AI Application Deployment and Inference, AI and DL Business Track (high level), Predictive Analytics for Retail, Consumer Engagement and Personalization
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9799
Streaming:
Download:
Share:
 
Abstract:
RAPIDS is an open-source platform for GPU data science, incubated by NVIDIA. Built to look and feel like popular tools in the Python Data Science ecosystem, RAPIDS is easy to use and dramatically speeds up execution of all steps of a typical dat ...Read More
Abstract:

RAPIDS is an open-source platform for GPU data science, incubated by NVIDIA. Built to look and feel like popular tools in the Python Data Science ecosystem, RAPIDS is easy to use and dramatically speeds up execution of all steps of a typical data science workflow. Intended for working data scientists, this session will be an in-depth walk through of all the stages of a model data science workflow using RAPIDS. The presentation will cover ingesting and cleaning data, feature engineering, working with strings, user-defined functions, and applying machine learning. The session will discuss the community and ecosystem around RAPIDS and future plans for the cuML library. Additionally, the session will cover how users can contribute to RAPIDS. At the end of the session, attendees will have learned RAPIDS benefits for data science, how to get started installing RAPIDS, and how to build their own workflows using RAPIDS.

  Back
 
Topics:
Accelerated Data Science, Algorithms and Numerical Techniques, RAPIDS
Type:
Tutorial
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9801
Streaming:
Download:
Share:
 
Abstract:
Traditional means of network mapping rely on expert knowledge, well-curated databases of network assets, and active internal scanning. Network maps are frequently out of date and often unable to provide the necessary ground-truth data to IT and secur ...Read More
Abstract:
Traditional means of network mapping rely on expert knowledge, well-curated databases of network assets, and active internal scanning. Network maps are frequently out of date and often unable to provide the necessary ground-truth data to IT and security. We'll show how to leverage RAPIDS and GPU-Accelerated data science to learn a network mapping from passively generated logs. We'll discuss how we take this a step further by applying multiple machine learning analytics to the graph to infer asset ownership, classify assets and services on the network, and provide near real-time updates and alerts based on changes to the network topology. We'll explain how near real-time ingest and processing capabilities allow us to visualize the network quickly and provide context to the security professional in a timely manner.  Back
 
Topics:
Accelerated Data Science, Cyber Security
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9802
Streaming:
Download:
Share:
 
Abstract:
Network defense and cybersecurity applications traditionally rely on heuristics and signatures to protect networks and detect anomalies. Large companies may generate over 10TB of data daily, spread across different sensors and heterogenous data ...Read More
Abstract:

Network defense and cybersecurity applications traditionally rely on heuristics and signatures to protect networks and detect anomalies. Large companies may generate over 10TB of data daily, spread across different sensors and heterogenous data types. The difficulty of providing timely ingest, feature engineering, feature exploration, and model generation has made signature-based detection the only option. We'll show how to use RAPIDS and GPU acceleration to overcome these obstacles. We'll walk through data engineering steps involving large amounts of heterogeneous data (both source and format) and explore how GPUs can accelerate feature exploration and hyperparameter selection. This enables more in-house data scientists and information security experts to use internally collected data to generate predictive models for anomaly detection rather than rely on signature-based detection.

  Back
 
Topics:
Accelerated Data Science, Cyber Security, RAPIDS
Type:
Tutorial
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9803
Streaming:
Download:
Share:
 
Abstract:
We'll discuss cuML, a GPU-Accelerated library of machine learning algorithms within the RAPIDS data science ecosystem. The cuML library allows data scientists, researchers, and software engineers to run traditional ML tasks on GPUs without g ...Read More
Abstract:

We'll discuss cuML, a GPU-Accelerated library of machine learning algorithms within the RAPIDS data science ecosystem. The cuML library allows data scientists, researchers, and software engineers to run traditional ML tasks on GPUs without going into the details of CUDA programming. We'll show you how to get tremendous speed-up for traditional machine learning workloads by using APIs like Scikit-Learn with Python. We'll also provide code examples, benchmarks, and the latest news.

  Back
 
Topics:
Accelerated Data Science, Tools and Libraries, RAPIDS
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9817
Streaming:
Download:
Share:
 
Abstract:
Releasing new feature updates of Windows 10 in a safe and efficient manner is of paramount importance to Microsoft. Each rollout goes through different phases (or rings), ensuring that we capture issues early and with limited impact on our customer b ...Read More
Abstract:
Releasing new feature updates of Windows 10 in a safe and efficient manner is of paramount importance to Microsoft. Each rollout goes through different phases (or rings), ensuring that we capture issues early and with limited impact on our customer base. However, the composition of features (hardware components and applications) in each ring is not truly representative of the retail population of PCs. Microsoft employees participate in testing of new features, and we release updates to a group of Windows Insiders and users seeking to check for updates. But he hardware and software features found in these populations do not provide us with full visibility into the ecosystem diversity of the retail population.  Back
 
Topics:
Accelerated Data Science
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9912
Streaming:
Download:
Share:
 
Abstract:
NVIDIA and HP are partnering to accelerate data science workloads. Hear from customers about how they are using Z by HP's data science workstation with NVIDIA RAPIDS technology to transform their workflows and learn how you could do the same ...Read More
Abstract:

NVIDIA and HP are partnering to accelerate data science workloads. Hear from customers about how they are using Z by HP's data science workstation with NVIDIA RAPIDS technology to transform their workflows and learn how you could do the same.

  Back
 
Topics:
Accelerated Data Science, RAPIDS
Type:
Sponsored Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9986
Streaming:
Download:
Share:
 
Abstract:
Learn how the OmniSci GPU-Accelerated SQL engine fits into the overall RAPIDS partner ecosystem for open source GPU analytics. Using open data, we'll show how to ingest data that's from both streaming and standing sources, perform descri ...Read More
Abstract:

Learn how the OmniSci GPU-Accelerated SQL engine fits into the overall RAPIDS partner ecosystem for open source GPU analytics. Using open data, we'll show how to ingest data that's from both streaming and standing sources, perform descriptive statistics and feature engineering using SQL and cuDF, and return the results as a GPU DataFrame. We'll also describe how data science workflow can be accomplished using tools from the RAPIDS ecosystem, all without the data ever leaving the GPU.

  Back
 
Topics:
Accelerated Data Science, Deep Learning and AI Frameworks, RAPIDS
Type:
Sponsored Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9992
Streaming:
Download:
Share:
 
Abstract:
Does everyone need billion-dollar investments and massive infrastructure to succeedin AI and machine learning? Or, could the answer be as simple as a better data science ecosystem and trusty, under-the-desk, workstations? This talk will focus on the ...Read More
Abstract:
Does everyone need billion-dollar investments and massive infrastructure to succeedin AI and machine learning? Or, could the answer be as simple as a better data science ecosystem and trusty, under-the-desk, workstations? This talk will focus on the state of data science workflow execution and how a workstation may just be the missing link to happier data scientists and increased productivity.  Back
 
Topics:
Accelerated Data Science, Tools and Libraries, AI and DL Business Track (high level)
Type:
Sponsored Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9996
Streaming:
Download:
Share:
 
Abstract:
Our talk covers architecture considerations for federating ML and DL data pipelines to exploit GPU acceleration for a seamless tier of data science deployments across edge, core, and cloud. We'll take a look at different requirements, architecture o ...Read More
Abstract:
Our talk covers architecture considerations for federating ML and DL data pipelines to exploit GPU acceleration for a seamless tier of data science deployments across edge, core, and cloud. We'll take a look at different requirements, architecture options to meet them, and the resulting benefits to deliver distributed deployments of the data pipeline stages across data ingestion, data prep, training, inference validation, data science, and model serving. We'll also explore a few ways in which customers are deploying these. We will be joined by implementers of stages of the AI and data pipeline today to hear about their deployments and experiences.  Back
 
Topics:
Accelerated Data Science, AI Application Deployment and Inference, Data Center and Cloud Infrastructure
Type:
Sponsored Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9997
Streaming:
Share:
Additive Manufacturing
Presentation
Media
Abstract:
For decades GPUs have blazed pixels to the screen, while 3D geometry kernels primarily ran on the CPU. What happens when you move up the stack, leveraging the GPU to create the geometry itself? We'll introduce our groundbreaking native GPU-Accelerat ...Read More
Abstract:
For decades GPUs have blazed pixels to the screen, while 3D geometry kernels primarily ran on the CPU. What happens when you move up the stack, leveraging the GPU to create the geometry itself? We'll introduce our groundbreaking native GPU-Accelerated geometry kernel, which is fully accessible via a Python API. We'll discuss the first application built on Dyndrite, the Additive Manufacturing Toolkit, which prints parts using the same splines used to design the parts. We'll also outline what's next for the software.  Back
 
Topics:
Additive Manufacturing, Digital Product Design & Styling
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9328
Streaming:
Download:
Share:
Advanced AI Learning Techniques
Presentation
Media
Abstract:
Generative Adversarial Networks (GANs) have produced amazing results in image generation, but they also have the potential to be applied to text and speech generation to overcome the limits of conventional methods. We'll introduce GANs and provide a ...Read More
Abstract:
Generative Adversarial Networks (GANs) have produced amazing results in image generation, but they also have the potential to be applied to text and speech generation to overcome the limits of conventional methods. We'll introduce GANs and provide a thorough review of this technology, then dive into how they can be used for speech signals and natural language processing.  Back
 
Topics:
Advanced AI Learning Techniques
Type:
Tutorial
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9127
Streaming:
Download:
Share:
 
Abstract:
We will discuss using spiking neural networks on GPUs to train audio and vision networks with our new BICHNN architecture. Spiking neural networks are more brain-like and have richer time-domain signal processing behavior than traditional feed-forwar ...Read More
Abstract:
We will discuss using spiking neural networks on GPUs to train audio and vision networks with our new BICHNN architecture. Spiking neural networks are more brain-like and have richer time-domain signal processing behavior than traditional feed-forward networks, but back propagation and gradient descent don't work on spiking neural nets. We'll describe how we train our models using our Bidirectional Interleaved Hierarchical Neural Networks. We will show how we construct these spiking neural networks in our NeuroCAD visual design tool, then connect them up with parameter-driven probability maps to give us the BICHNN architecture. Then we will turn them loose and train them in real-time and demonstrate how we use genetic algorithms and massive amounts of GPU simulation time to optimize the networks to the specified task.  Back
 
Topics:
Advanced AI Learning Techniques, Deep Learning and AI Frameworks
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9151
Streaming:
Download:
Share:
 
Abstract:
This presentation illustrates how to combine innovations from several sub-disciplines of machine learning research to train understandable, fair, trustable, and accurate predictive modeling systems. Techniques from research into fair models, directly ...Read More
Abstract:
This presentation illustrates how to combine innovations from several sub-disciplines of machine learning research to train understandable, fair, trustable, and accurate predictive modeling systems. Techniques from research into fair models, directly interpretable Bayesian or constrained machine learning models, and post-hoc explanations can be used to train transparent, fair, and accurate models and make nearly every aspect of their behavior understandable and accountable to human users. Additional techniques from fairness research can be used to check for sociological bias in model predictions and to preprocess data and post-process predictions to ensure the fairness of predictive models. Finally, applying new testing and debugging techniques, often inspired by best practices in software engineering, can increase the trustworthiness of model predictions on unseen data. Together these techniques create a new and truly human-friendly type of machine learning suitable for use in business- and life-critical decision support.  Back
 
Topics:
Advanced AI Learning Techniques, Deep Learning and AI Frameworks
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9249
Streaming:
Download:
Share:
 
Abstract:
We'll explore how to discover properties of deep networks by looking at their learned parameters or measuring the patterns of the networks' input space. Emerging properties from individual samples can be measured by examining the common changes the ...Read More
Abstract:
We'll explore how to discover properties of deep networks by looking at their learned parameters or measuring the patterns of the networks' input space. Emerging properties from individual samples can be measured by examining the common changes they undergo during training. We'll explain how this allows a hierarchical analysis that goes beyond explainability of individual decisions why a particular image was misclassified, for example and extends to entire classes or even the training dataset itself. We show how understanding these patterns can provide the foundation for more principled, stable, and robust definitions of future network architectures and more consistent learning procedures.  Back
 
Topics:
Advanced AI Learning Techniques, Computer Vision
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9287
Streaming:
Download:
Share:
 
Abstract:
Learn how the combination of capsule networks, active learning, and transfer learning can reduce the number of training samples required to add a new label to an existing classifier. We will detail how we developed our network architecture, training ...Read More
Abstract:
Learn how the combination of capsule networks, active learning, and transfer learning can reduce the number of training samples required to add a new label to an existing classifier. We will detail how we developed our network architecture, training data selection algorithm, and discuss their implementation in Python and TensorFlow's Keras layers with GPU acceleration. We'll also discuss our results from applying this approach to image-classification tasks and how they compare to a standard convolutional neural network approach.  Back
 
Topics:
Advanced AI Learning Techniques, AI and DL Research, Computer Vision
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9290
Streaming:
Download:
Share:
 
Abstract:
Learn how our Esri GeoAI team is working to increase security and reduce crime by developing optimal police patrol strategies. We'll describe our work at the intersection of deep reinforcement learning and spatial statistics, and explain how we're ...Read More
Abstract:
Learn how our Esri GeoAI team is working to increase security and reduce crime by developing optimal police patrol strategies. We'll describe our work at the intersection of deep reinforcement learning and spatial statistics, and explain how we're applying statistical models to produce an environment for training a police-patrol dispatch and control agent. The massive compute for training the agent comes from intelligent application of distributed GPU training. We'll detail our agent model and distributed training, and explain how spatial statistics and GIS are needed to solve this challenging problem.  Back
 
Topics:
Advanced AI Learning Techniques, AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9294
Streaming:
Download:
Share:
 
Abstract:
We'll discuss monitoring and visualizing a deep neural network in MXNet and explain how to improve training performance. We'll also talk about coding best practices, data pre-processing, making effective use of CPUs, hybridization, efficient batch ...Read More
Abstract:
We'll discuss monitoring and visualizing a deep neural network in MXNet and explain how to improve training performance. We'll also talk about coding best practices, data pre-processing, making effective use of CPUs, hybridization, efficient batch size, low precision training, and other tips and tricks that can improve training performance by orders of magnitude.  Back
 
Topics:
Advanced AI Learning Techniques
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9370
Streaming:
Download:
Share:
 
Abstract:
Learn about deep active learning and how it's been successfully applied to the autonomous vehicle project at NVIDIA. We'll discuss aspects of our work on learning from limited labels for autonomous vehicle and other datasets, and outline the import ...Read More
Abstract:
Learn about deep active learning and how it's been successfully applied to the autonomous vehicle project at NVIDIA. We'll discuss aspects of our work on learning from limited labels for autonomous vehicle and other datasets, and outline the importance of advanced data-driven methods to build datasets. We'll also describe our plans for research in learning from limited labels and optimal construction of datasets. In addition, our talk will touch on more recent and exploratory experiments toward optimal construction of training sets for today's deep neural network models and other machine learning and AI methods.  Back
 
Topics:
Advanced AI Learning Techniques, Autonomous Vehicles
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9613
Streaming:
Download:
Share:
 
Abstract:
Learn how to reconstruct missing data in specific city areas when only one source of data is available, and find out how to predict activity patterns based on historical urban data. We'll explain how we achieve this by training a neural network to l ...Read More
Abstract:
Learn how to reconstruct missing data in specific city areas when only one source of data is available, and find out how to predict activity patterns based on historical urban data. We'll explain how we achieve this by training a neural network to learn the relationship between different types of data such as spend data and activity data. We used this approach to predict the local economic impact of building a new community center or holding a large event. We'll discuss how we use the power of modern NVIDIA GPUs to run simulations of whole cities and how we use Chronotope visualization software to let analysts explore predictions and compare these with existing situations.  Back
 
Topics:
Advanced AI Learning Techniques, Deep Learning and AI Frameworks
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9660
Streaming:
Download:
Share:
 
Abstract:
Although some fear AI will assume human jobs and lead to mass unemployment, we argue that the most creative and productive outcomes will occur when humans and machines work together to enhance complementary strengths and skills. By augmenting human c ...Read More
Abstract:
Although some fear AI will assume human jobs and lead to mass unemployment, we argue that the most creative and productive outcomes will occur when humans and machines work together to enhance complementary strengths and skills. By augmenting human capabilities and pushing the boundaries of creativity, can AI help us create things that wouldn't have existed otherwise? We'll describe our experiments to design haute couture dresses, cook pizza with shrimp and jam, create a scent, paint the walls of Graffiti Alley in Cambridge, and drink cocktails, all inspired by AI-generated content. Learn how to generate new objects and experiences with AI and how to boost your business with AI-augmented creativity.  Back
 
Topics:
Advanced AI Learning Techniques, AI Application Deployment and Inference
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9835
Streaming:
Download:
Share:
 
Abstract:
Learn about a new way to generate photorealistic videos using high-level representations of a scene such as semantic maps, edges, or keypoints. This makes synthesizing new videos much easier and more intuitive. We'll show that this has potential app ...Read More
Abstract:
Learn about a new way to generate photorealistic videos using high-level representations of a scene such as semantic maps, edges, or keypoints. This makes synthesizing new videos much easier and more intuitive. We'll show that this has potential applications in many areas. For example, it can replace a traditional graphics pipeline with AI-based rendering. It can also manipulate existing videos in ways that include swapping faces of a talking person or transferring motions from one person to another.  Back
 
Topics:
Advanced AI Learning Techniques, Graphics and AI
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9904
Streaming:
Download:
Share:
Algorithms and Numerical Techniques
Presentation
Media
Abstract:
Learn about using Tensor Cores to perform very fast matrix multiply-accumulate steps like those required in AI training. The key to Tensor Core performance is the use of 16-bit floating point arithmetic, but that causes significant rounding erro ...Read More
Abstract:

Learn about using Tensor Cores to perform very fast matrix multiply-accumulate steps like those required in AI training. The key to Tensor Core performance is the use of 16-bit floating point arithmetic, but that causes significant rounding errors. Although algorithms like binomial correction or Karatsuba can reduce rounding errors considerably, they require additional calculations. We'll detail performance of these algorithms based on the Warp Matrix Multiply Accumulate API.

  Back
 
Topics:
Algorithms and Numerical Techniques, HPC and AI
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9176
Streaming:
Download:
Share:
 
Abstract:
<div> Learn how to implement state-of-the-art preconditioners for iterative solvers of large-scale linear systems in CUDA. Previously most preconditioners were set up on CPUs because this task was not considered suitable for fine-grain parallel ...Read More
Abstract:
<div> Learn how to implement state-of-the-art preconditioners for iterative solvers of large-scale linear systems in CUDA. Previously most preconditioners were set up on CPUs because this task was not considered suitable for fine-grain parallelization. We&#39;ll show how it&#39;s possible to implement efficient CUDA kernels for techniques like the adaptive factorized sparse approximate inverse by adopting an approach that dramatically reduces the amount of memory required to run in parallel. We&#39;ll describe how our GPU-only preconditioners and solvers can be used to solve real-world problems in science and engineering. We&#39;ll provide single and multi-GPU implementations. Our method makes it possible to obtain about an order-of-magnitude speedup on high-end multi-core CPUs like the Intel Xeon Platinum 8176.</div> <div> &nbsp;</div>  Back
 
Topics:
Algorithms and Numerical Techniques, Tools and Libraries
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9192
Streaming:
Download:
Share:
 
Abstract:
We'll discuss an approximate singular value decomposition that is much faster than state-of-the-art SVD and maintains the same accuracy if the requested singular values are away from zero. Learn how to trade off performance and accuracy during this ...Read More
Abstract:
We'll discuss an approximate singular value decomposition that is much faster than state-of-the-art SVD and maintains the same accuracy if the requested singular values are away from zero. Learn how to trade off performance and accuracy during this talk.  Back
 
Topics:
Algorithms and Numerical Techniques, Accelerated Data Science
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9226
Streaming:
Download:
Share:
 
Abstract:
The Volta architecture's comprehensive support for synchronization results from independent thread scheduling. We'll provide an in-depth analysis of independent thread scheduling. Our talk will cover situations in which device-side synchronization ...Read More
Abstract:
The Volta architecture's comprehensive support for synchronization results from independent thread scheduling. We'll provide an in-depth analysis of independent thread scheduling. Our talk will cover situations in which device-side synchronization is useful, clarify what is possible, and propose a starting point for best practices. Where possible, we'll also retrace the steps that led to the best practices outside of accelerated computing, and emphasize what has changed. The only prerequisite to attend this talk is to have some experience with accelerated programming, although even expert computer scientists will learn something new.  Back
 
Topics:
Algorithms and Numerical Techniques, Programming Languages
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9329
Streaming:
Download:
Share:
 
Abstract:
Learn how to find k-cores in graphs efficiently on GPUs using dynamic graph operations. The k-core of a graph is a metric used in social networks analytics, visualization, graph coloring, and other applications. We'll discuss a new parallel and scal ...Read More
Abstract:
Learn how to find k-cores in graphs efficiently on GPUs using dynamic graph operations. The k-core of a graph is a metric used in social networks analytics, visualization, graph coloring, and other applications. We'll discuss a new parallel and scalable algorithm for finding the maximal k-core implemented. When run on an NVIDIA Tesla P100, that process is up to 58x faster than a sequential graph implementation and up to 4x faster than a similar parallel algorithm on a 36-core CPU. We'll explain how to extend our algorithm to support k-core edge decomposition for different size k-cores found in the graph. Our k-core decomposition algorithm on the P100 is up to 130x faster than sequential graph and up to 8x faster than the same CPU-based parallel algorithm. We'll also show how our algorithm finds a k-core with dynamic graph operations rather using a static graph.  Back
 
Topics:
Algorithms and Numerical Techniques, In-Situ and Scientific Visualization
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9440
Streaming:
Download:
Share:
Animation and VFX
Presentation
Media
Abstract:
RED will discuss how its R3D SDK can be used to make 8K workflow effortless, including up to 10x faster transcoding and real-time playback performance of 8K digital footage within applications. We'll discuss some of the challenges RED engineers face ...Read More
Abstract:
RED will discuss how its R3D SDK can be used to make 8K workflow effortless, including up to 10x faster transcoding and real-time playback performance of 8K digital footage within applications. We'll discuss some of the challenges RED engineers faced in achieving real-time and how NVIDIA's technologies enabled us to overcome those challenges. We will also walk through the process of integrating RED's R3D SDK into your application to take full advantage of the improved image pipeline for processing 8K footage.&nbsp;  Back
 
Topics:
Animation and VFX
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S91041
Streaming:
Download:
Share:
 
Abstract:
Legacy production methods can't keep up with the global nature of content creation. Studios need to operate where tax incentives are offered, and artist talent may be located anywhere in the world. A Virtual Studio lets you deploy resources ...Read More
Abstract:

Legacy production methods can't keep up with the global nature of content creation. Studios need to operate where tax incentives are offered, and artist talent may be located anywhere in the world. A Virtual Studio lets you deploy resources where and when you need them in a matter of minutes, rather than weeks, so you can ramp up and down as production ebbs and flows. Virtual Studios are fueled by GPUs, which provide artists and engineers with both a powerful virtual workstation and the ability to accelerate renders and simulations, both locally and distributed across clusters. On the cloud, you're able to visualize and manipulate massive datasets that would be difficult or even impossible to achieve on traditional hardware. This session examines the benefits, strategies, and challenges of building a Virtual Studio on Google Cloud Platform, powered by NVIDIA GPUs.

  Back
 
Topics:
Animation and VFX
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S91042
Streaming:
Share:
 
Abstract:
We'll discuss Genesis, MPC's virtual production platform, designed as a robust multi-user distributed system that incorporates both modern technologies like mixed reality and more traditional techniques like motion capture and camera operation via ...Read More
Abstract:
We'll discuss Genesis, MPC's virtual production platform, designed as a robust multi-user distributed system that incorporates both modern technologies like mixed reality and more traditional techniques like motion capture and camera operation via encoded hardware devices. We'll talk about how we're improving the quality of the real-time graphics, with a special attention to lighting. We will explain how we have started incorporating elements of real-time ray tracing into our platform, from a live link to Renderman XPU and our own Optix-based path tracer, as well as a hybrid approach based on DXR running inside Unity.  Back
 
Topics:
Animation and VFX, Rendering and Ray Tracing
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9132
Streaming:
Download:
Share:
 
Abstract:
We'll discuss a deep-learning approach that takes as an input a few images of a scene and synthesizes new views as seen from virtual cameras. This could be used to generate videos such as camera flyby videos, or simply a view of the scene from a new ...Read More
Abstract:
We'll discuss a deep-learning approach that takes as an input a few images of a scene and synthesizes new views as seen from virtual cameras. This could be used to generate videos such as camera flyby videos, or simply a view of the scene from a new location. Despite several novel view synthesis approaches, the quality of resulting images quickly degrades when the virtual camera moves significantly with respect to the input images due to increasing depth uncertainty and disocclusions. We'll describe how we cast this problem as one of depth probability estimation for the novel view, image synthesis, and conditional image refinement. We'll also cover traditional and deep learning-based depth estimation, issues with warping-based novel view synthesis methods, and how depth information can be used to refine the quality of synthesized images.  Back
 
Topics:
Animation and VFX, Computer Vision, Video and Image Processing
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9576
Streaming:
Share:
 
Abstract:
We'll examine the potential for spatial computing and machine learning to reintroduce people to the physical potential of their bodies by focusing on Embody, MAP Lab's 2019 Sundance premiere. Inspired by movement traditions such as aikid ...Read More
Abstract:

We'll examine the potential for spatial computing and machine learning to reintroduce people to the physical potential of their bodies by focusing on Embody, MAP Lab's 2019 Sundance premiere. Inspired by movement traditions such as aikido, yoga, and dance, Embody is a social VR experience that uses visual metaphor and encouragement from teachers and friends to bring about coordinated body movement. We'll explain how this experience, which is piloted entirely by body movement and position, reclaims the body's potential inside the digital landscape. Users prompt each other with conversation, mirroring, and environmental channeling to step together through physical sequences designed to center, balance, extend, and strengthen. We hope players who experience Embody will be reminded of their deep physical potential and remember that the body is a flexible tool and able to change (http://www.sundance.org/projects/embody).

  Back
 
Topics:
Animation and VFX, Virtual Reality and Augmented Reality, AI for Gaming
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9903
Streaming:
Download:
Share:
 
Abstract:
During this session you will learn how Digital Domain and Nvidia create compelling digital characters that are used for feature films and real-time performance events. You will learn the history of some digital characters from our library and see com ...Read More
Abstract:
During this session you will learn how Digital Domain and Nvidia create compelling digital characters that are used for feature films and real-time performance events. You will learn the history of some digital characters from our library and see compelling behind the scenes footage. From Tupac at Coachella to the machine learning algorithms that helped drive Thanos, Digital Domain is pioneering how we use machine learning to drive real-time performances, both on-stage and off.  Back
 
Topics:
Animation and VFX
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9910
Streaming:
Download:
Share:
 
Abstract:
Generative methods allow a computer to automatically distill the essence of a dataset and then produce novel examples that are indistinguishable from the original data. That's the promise, but getting there has been difficult. This talk focu ...Read More
Abstract:

Generative methods allow a computer to automatically distill the essence of a dataset and then produce novel examples that are indistinguishable from the original data. That's the promise, but getting there has been difficult. This talk focuses on recent advances in generative adversarial networks (GAN), describing ideas that have finally enabled the synthesis of credible high-resolution images. It also covers recent work by NVIDIA (StyleGAN) that makes the image generation more controllable by borrowing ideas from style transfer literature, and also leads to an interesting, unsupervised separation of high-level attributes (e.g. pose or identity in case of human faces) and inconsequential variation in the images (exact placement of hair, etc.).

  Back
 
Topics:
Animation and VFX, AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9952
Streaming:
Share:
Astronomy and Astrophysics
Presentation
Media
Abstract:
After the Big Bang, the Universe contained hydrogen, helium, and a bit of lithium. Every other element on the periodic table is produced in stars and is disseminated into interstellar space via supernova explosions. Simulations of supernovae are amon ...Read More
Abstract:
After the Big Bang, the Universe contained hydrogen, helium, and a bit of lithium. Every other element on the periodic table is produced in stars and is disseminated into interstellar space via supernova explosions. Simulations of supernovae are among the most compute-intensive multi-physics applications on the world's largest modern supercomputers. We will discuss recent development of the FLASH code intended to make these simulations even more physically meaningful. In particular, well describe how our work on FLASH, as part of the OLCF CAAR program, allowed us to increase the number of tracked nuclear species from about a dozen to hundreds, making precision predictions that can be compared to observations possible.  Back
 
Topics:
Astronomy and Astrophysics, Computational Physics
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S91009
Streaming:
Share:
 
Abstract:
The Square Kilometre Array is a planned next-generation radio telescope. It will be used to answer fundamental questions such as what is dark energy and dark matter? Is Einstein's theory of general relativity correct? And, are we alone in the univer ...Read More
Abstract:
The Square Kilometre Array is a planned next-generation radio telescope. It will be used to answer fundamental questions such as what is dark energy and dark matter? Is Einstein's theory of general relativity correct? And, are we alone in the universe? To answer such questions the telescope must collect vast amounts of data. This data needs to pass through complex signal processing pipelines for science products to be extracted. This talk will introduce SKA and the science it aims to achieve and discuss how GPUs can be used to achieve this.We'll discuss current advances in the AstroAccelerate software package, which is GPU-Enabled and written in CUDA. AstroAccelerate focuses on enabling real-time processing of time-domain radio-astronomy data.  Back
 
Topics:
Astronomy and Astrophysics
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9286
Streaming:
Download:
Share:
 
Abstract:
We'll provide an overview of the Breakthrough Listen Initiative launched in 2015 by the Royal Society in London to conduct the most comprehensive and sensitive search for extraterrestrial intelligence in history. The search, characterized by high da ...Read More
Abstract:
We'll provide an overview of the Breakthrough Listen Initiative launched in 2015 by the Royal Society in London to conduct the most comprehensive and sensitive search for extraterrestrial intelligence in history. The search, characterized by high data volume and complex interference environments, needs advanced methods of artificial intelligence. We will discuss the project's NVIDIA GPU-Powered digital backend, with a focus on data analysis techniques and deep learning applications. Highlights include applications of predictive anomaly detections, supervised and semi-supervised signal classifications, and detections of fast radio bursts, mysterious signals of unclear origin from billions of light years away.  Back
 
Topics:
Astronomy and Astrophysics, AI and DL Research, AI Application Deployment and Inference
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9307
Streaming:
Download:
Share:
 
Abstract:
AI and related technologies are beginning to revolutionize astronomy and astrophysics. As facilities like the Large Synoptic Survey Telescope and the Wide Field InfraRed Telescope come online, data volumes in astronomy will increase. We will describe ...Read More
Abstract:
AI and related technologies are beginning to revolutionize astronomy and astrophysics. As facilities like the Large Synoptic Survey Telescope and the Wide Field InfraRed Telescope come online, data volumes in astronomy will increase. We will describe a deep learning framework that allows astronomers to identify and categorize astronomical objects in enormous datasets with more fidelity than ever. We'll also review new applications of AI in astrophysics, including data analysis and numerical simulation.  Back
 
Topics:
Astronomy and Astrophysics, Deep Learning and AI Frameworks, HPC and AI
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9508
Streaming:
Download:
Share:
 
Abstract:
Learn how GPUs are pushing the limits of the largest astronomical telescopes on Earth and how they'll be used to image life-bearing planets outside our solar system. Thanks to hardware features such as Tensor Cores and mixed-precision suppor ...Read More
Abstract:

Learn how GPUs are pushing the limits of the largest astronomical telescopes on Earth and how they'll be used to image life-bearing planets outside our solar system. Thanks to hardware features such as Tensor Cores and mixed-precision support, plus optimized AI frameworks, GPU technology is changing how large data streams from optical sensors are digested in real time. We'll discuss how real-time AI made possible by GPUs opens up new means to optimally control the system and calibrate images, which will help scientists get the most out of the largest optical telescopes. GPUs will also benefit future extreme-size facilities like the European Extremely Large Telescope because the complexity of maintaining exquisite image quality increases with the square of its diameter size. We'll present on-sky results obtained on the 8.2-meter Subaru Telescope and explain why these techniques will be essential to future giant telescopes.

  Back
 
Topics:
Astronomy and Astrophysics, HPC and AI
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9634
Streaming:
Download:
Share:
 
Abstract:
Learn about Cholla, a GPU-Native massively parallel hydrodynamics code that runs on the world's largest supercomputers and is pushing the forefront of astrophysical research. We'll describe Cholla's design, including our innovations in transferrin ...Read More
Abstract:
Learn about Cholla, a GPU-Native massively parallel hydrodynamics code that runs on the world's largest supercomputers and is pushing the forefront of astrophysical research. We'll describe Cholla's design, including our innovations in transferring classic computational fluid dynamics algorithms to GPUs. We'll cover our ongoing research in astrophysics, highlighting results from our 2017-2018 INCITE program to understand the role galactic winds play in the ongoing evolution of galaxies. In addition, we'll describe our current efforts to couple Cholla with NVIDIA's IndeX visualization software to provide high-fidelity in-situ renderings of our simulations.  Back
 
Topics:
Astronomy and Astrophysics, HPC and Supercomputing
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9728
Streaming:
Download:
Share:
Autonomous Vehicles
Presentation
Media
Abstract:
The auto industry is moving at a rapid pace toward autonomous vehicles. While some players are focusing on bringing fully driverless robotaxis to market, with massive sensing and computing power at high cost, many automakers are looking into how to e ...Read More
Abstract:
The auto industry is moving at a rapid pace toward autonomous vehicles. While some players are focusing on bringing fully driverless robotaxis to market, with massive sensing and computing power at high cost, many automakers are looking into how to expand the current driver assistance (ADAS) systems to partial (SAE Level 2) and/or conditional automation (SAE Level 3). These companies are looking to deploy in the next few years, within the constraints of limited compute resources and low cost sensors cameras, radars, and low-cost GPS receivers. In this talk, we will dissect the key technology and system challenges of enabling autonomy within these constraints, and how XPeng Motors plans to solve these challenges in the China market.  Back
 
Topics:
Autonomous Vehicles
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S91049
Streaming:
Share:
 
Abstract:
The autonomous driving industry is busy solving the challenging and complex algorithm development for perception and path planning. This presentation will focus on the next, rarely discussed challenge: building a distributed system to run these algor ...Read More
Abstract:
The autonomous driving industry is busy solving the challenging and complex algorithm development for perception and path planning. This presentation will focus on the next, rarely discussed challenge: building a distributed system to run these algorithms, made up of many processors, accelerators, switches, and all of the periphery hardware. It will also explore guarantees about performance, determinism, bandwidth, failures, and safety.&nbsp;  Back
 
Topics:
Autonomous Vehicles
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S91050
Streaming:
Download:
Share:
 
Abstract:
Learn about Toyota's research on stereo-based deep learning for environment perception in challenging environments. We'll describe our novel ChiNet deep learning framework, implemented on a GPU for effective sensor fusion. We'll explain how we per ...Read More
Abstract:
Learn about Toyota's research on stereo-based deep learning for environment perception in challenging environments. We'll describe our novel ChiNet deep learning framework, implemented on a GPU for effective sensor fusion. We'll explain how we perform free space and road object estimation in a variety of environments by using a thermal stereo pair to estimate obstacle depth. The session will highlight our results, which show that the sensor fusion framework improves robustness and accuracy of deep learning-based free space object detection. We'll also cover our plan to implement the framework on the NVIDIA DRIVE AGX Pegasus platform.  Back
 
Topics:
Autonomous Vehicles, AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9195
Streaming:
Download:
Share:
 
Abstract:
The Toyota Research Institute is going beyond supervised learning for automated driving and exploring problems that affect research and development of long-term, large-scale autonomous robots. These problems include unsupervised domain adaptation, se ...Read More
Abstract:
The Toyota Research Institute is going beyond supervised learning for automated driving and exploring problems that affect research and development of long-term, large-scale autonomous robots. These problems include unsupervised domain adaptation, self-supervised learning, and robustness to edge cases. This session will dive into robotics systems, especially end-to-end vs. modular design and human-robot interaction. It will also include some of TRI's related research directions, especially those around world-scale cloud robotics.  Back
 
Topics:
Autonomous Vehicles, AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9314
Streaming:
Download:
Share:
 
Abstract:
Leaders from the mapping technology companies will discuss the advantages of various algorithms to create and maintain maps, followed by a short Q&A session. HERE: Vladimir Shestak, Lead Software Engineer Automated Driving Edge Perception fo ...Read More
Abstract:

Leaders from the mapping technology companies will discuss the advantages of various algorithms to create and maintain maps, followed by a short Q&A session. HERE: Vladimir Shestak, Lead Software Engineer Automated Driving Edge Perception for HD Map Maintenance: We start this talk by presenting a brief overview of HD Live Map created by HERE and its use for connected ADAS or automated driving solutions. Although building such a map with a required centimeter level precision is technically hard, the instant the HD Live Map is built, changes in the real world can occur causing the map to no longer reflect reality.  Hence, a proper maintenance strategy must be in place with the goal to identify discrepancies between the HD Live Map and the real world and heal the HD Live Map as quickly as possible. We discuss a spectrum of techniques developed by HERE to address the map-healing process and then focus on our low-cost solutions for in-vehicle change detection. The example system employs a consumer-grade Android-based sensing system streaming imagery and telemetry in real-time into HERE Edge Perception software stack. We present the high-level software architecture of the stack, its main components, i.e., feature detection, object tracking and triangulation, RWO and Maplet generation, as well as in-vehicle deployment options. The real-time performance evaluation of the system concludes our talk. NavInfo Europe:  Geetank Raipuria, Computer Vision Engineer Real-Time Object Detection and Semantic Segmentation: This session will discuss how NavInfo uses computer vision and deep learning to build high-definition maps that cover China's highways and large city streets. This involves performing object detection and semantic segmentation on visual imagery collected from vehicle sensors. The NavInfo Europe Advanced Research Lab creates processes that extract information from this data, both in real-time onboard vehicles using the NVIDIA DRIVE platform, and faster than real-time, processing offline gathered video material through NVIDIA DeepStream.

  Back
 
Topics:
Autonomous Vehicles, Computer Vision
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9351
Streaming:
Download:
Share:
 
Abstract:
Autonomous vehicles will require in-cabin sensing to respond to user activities and emotional states. This session will provide an overview of Affectiva's Emotion AI technology to redefine the in-cabin experience. We will describe the progression of ...Read More
Abstract:
Autonomous vehicles will require in-cabin sensing to respond to user activities and emotional states. This session will provide an overview of Affectiva's Emotion AI technology to redefine the in-cabin experience. We will describe the progression of unimodal analysis of face and voice emotions to the fusion of these modalities to detect affective states, such as frustration. This process employs techniques such as deep learning-based spatio-temporal modeling, in addition to large-scale natural datasets -- made larger still by cross-domain augmentation -- to develop AI systems that can detect these affective states.  Back
 
Topics:
Autonomous Vehicles, Deep Learning and AI Frameworks, AI Application Deployment and Inference
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9466
Streaming:
Download:
Share:
 
Abstract:
Its time to separate the signal from the noise when it comes to autonomous driving. And as self-driving trucks near commercial reality, the stakes are high for safe operation on our highways. Join Dr. Xiaodi Hou, Founder, Pre ...Read More
Abstract:

Its time to separate the signal from the noise when it comes to autonomous driving. And as self-driving trucks near commercial reality, the stakes are high for safe operation on our highways. Join Dr. Xiaodi Hou, Founder, President and CTO of TuSimple, the largest self-driving truck company worldwide, for a discussion of what it takes to design, test and deploy a fully autonomous truck. Dr. Hou will lay it on the line in terms of whats working and whats not in the design and testing of todays self-driving trucks.

  Back
 
Topics:
Autonomous Vehicles, Intelligent Machines and IoT
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9555
Streaming:
Download:
Share:
 
Abstract:
As deep learning algorithms for autonomous driving have progressed from early semantic segmentation to today's advanced systems, their consumption of data and computational resources has increased. The amount of data acquired and the need for annota ...Read More
Abstract:
As deep learning algorithms for autonomous driving have progressed from early semantic segmentation to today's advanced systems, their consumption of data and computational resources has increased. The amount of data acquired and the need for annotations are growing exponentially, creating new challenges for improving accuracy and achieving desired safety levels. This session will explore some of these hurdles and discuss our proposed solutions in terms of active learning, computational efficiency, and the efficient use of synthetic data for training deep neural networks.  Back
 
Topics:
Autonomous Vehicles, AI and DL Research, Computer Vision
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9630
Streaming:
Download:
Share:
 
Abstract:
Moderator: Dr. Justyna Zander, Global Head of Mapping, NVIDIA This session will discuss NVIDIA DRIVE Mapping, a platform that enables vehicle manufacturers to use maps from various global providers for highly accurate navigation and localization. DRI ...Read More
Abstract:
Moderator: Dr. Justyna Zander, Global Head of Mapping, NVIDIA This session will discuss NVIDIA DRIVE Mapping, a platform that enables vehicle manufacturers to use maps from various global providers for highly accurate navigation and localization. DRIVE Mapping products integrate a scalable sensor suite, software development kits, and co-integrated high-definition maps from leading mapping companies. These end-to-end technologies help collect environmental data to create and update HD maps. We'll explain how the platform makes it possible for a self-driving vehicle to localize itself with precision, discern potential hazards, and determine exactly where it can safely drive. Leaders from the mapping technology companies will discuss the advantages of various modalities of maps and the benefit they provide to autonomous vehicles, followed by a short Q&amp;A session. TomTom: Willem Strijbosch, Head of Autonomous Driving Mapping Progress on the Car-to-Cloud-to-Car Cycle The talk will discuss the latest on map creation and using crowdsourced data for map updates at TomTom.&nbsp; 3DMapping: Dr. Gunnar Gr&auml;fe, CEO and Founder Precise Ultra HD Map Data as Basis for Virtual Testing and Simulation Digital road data is the basis for virtual testing and simulation. Artificially designed digital roads may help case by case, but for various applications the precise digitalization and digital as-built representation of real-world roads is needed. The typical requirement is, that the roads used for virtual testing and simulation are regarded as digital twin of the real-world roads, which is prerequisite for comparable testing in reality and in the virtual environment. The technical solution for digitizing test tracks, race tracks and public roads with sufficient accuracy and resolution is high-end mobile surveying using high-resolution scanners and multiple cameras. 3D Mapping has invented the necessary technology since more than 20 years and today deploys van-based survey systems worldwide. The technology is used for example to generate high-resolution digital road surface models in OpenCRG format or to produce precise high definition reference maps in OpenDrive format, which are either used for virtual simulation and testing or as reference map in the car for autonomous driving development. 3D Mapping is member of the OpenDrive core team and has been intensively working on standardization and updates of the formats OpenDrive and OpenCRG since several years and is fully engaged in the ongoing ASAM format standardizations. The developments lead to new standards including 3D environment combined with scenario elements.  Back
 
Topics:
Autonomous Vehicles
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9771
Streaming:
Download:
Share:
 
Abstract:
We will elaborate on how our holistic approach to design and validation creates a single environment to engineer and experience the autonomous vehicle. From Cognitive Augmented Design & Model Based System Engineering to realistic validation ...Read More
Abstract:

We will elaborate on how our holistic approach to design and validation creates a single environment to engineer and experience the autonomous vehicle. From Cognitive Augmented Design & Model Based System Engineering to realistic validation at scale, AI is enabling AV developers to increase safety while managing the costs of ever-increasing complexity.  

  Back
 
Topics:
Autonomous Vehicles
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9773
Streaming:
Download:
Share:
 
Abstract:
Automated driving systems are challenging to bring to market due to an enormous number of scenarios and environment parameter combinations that must be validated. Using current real-world tests for human-driven vehicles on newly developed automated d ...Read More
Abstract:
Automated driving systems are challenging to bring to market due to an enormous number of scenarios and environment parameter combinations that must be validated. Using current real-world tests for human-driven vehicles on newly developed automated driving technology is no longer feasible. This session will explore how testing in the virtual world enables manufacturers and regulators to validate a wide variety of traffic situations without hazard. Especially when using deep learning algorithms for automated driving functions, a scalable, powerful, and consistent toolchain is required. To homologate automated vehicles with confidence, simulation platforms will be crucial.  Back
 
Topics:
Autonomous Vehicles
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9804
Streaming:
Download:
Share:
 
Abstract:
One of the most exciting aspects of working on autonomous vehicles is experiencing self-driving car functionality live in the car. This session will share an inside look into NVIDIA's most advanced autonomous vehicle drive missions, powered by the D ...Read More
Abstract:
One of the most exciting aspects of working on autonomous vehicles is experiencing self-driving car functionality live in the car. This session will share an inside look into NVIDIA's most advanced autonomous vehicle drive missions, powered by the DRIVE platform.  Back
 
Topics:
Autonomous Vehicles
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9811
Streaming:
Download:
Share:
 
Abstract:
This session will discuss the process of training deep neural networks using NVIDIA DGX servers at BMW Group. We will describe our research work in four application areas: fine-grained vehicle representations for autonomous driving, panoptic segmenta ...Read More
Abstract:
This session will discuss the process of training deep neural networks using NVIDIA DGX servers at BMW Group. We will describe our research work in four application areas: fine-grained vehicle representations for autonomous driving, panoptic segmentation, self-supervised learning of the drivable area for autonomous vehicles and neural network optimization. All of these projects require high-performance compute and demand a scalable, agile and adaptive learning infrastructure, leveraging Kubernetes on NVIDIA DGX servers.  Back
 
Topics:
Autonomous Vehicles, AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9892
Streaming:
Download:
Share:
 
Abstract:
Autonomous driving systems use various neural network models that require extremely accurate and efficient computation on GPUs. This session will outline how Zoox employs two strategies to improve inference performance (i.e., latency) of trained neur ...Read More
Abstract:
Autonomous driving systems use various neural network models that require extremely accurate and efficient computation on GPUs. This session will outline how Zoox employs two strategies to improve inference performance (i.e., latency) of trained neural network models without loss of accuracy: (1) inference with NVIDIA TensorRT, and (2) inference with lower precision (i.e., Fp16 and Int8). We will share our learned lessons about neural network deployment with TensorRT and our current conversion workflow to tackle limitations.  Back
 
Topics:
Autonomous Vehicles
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9895
Streaming:
Download:
Share:
 
Abstract:
NVIDIA DRIVE Constellation is a virtual reality simulation platform designed to support the development and validation of autonomous vehicles. It is a cloud-based platform that enables hardware-in-the-loop testing and large-scale deployment in data c ...Read More
Abstract:
NVIDIA DRIVE Constellation is a virtual reality simulation platform designed to support the development and validation of autonomous vehicles. It is a cloud-based platform that enables hardware-in-the-loop testing and large-scale deployment in data centers, and is capable of driving millions of testing miles. We'll discuss how NVIDIA DRIVE Constellation is being used for the validation of safe autonomous driving. We will also discuss how companies can partner with NVIDIA and join the DRIVE Constellation ecosystem.  Back
 
Topics:
Autonomous Vehicles
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9897
Streaming:
Share:
 
Abstract:
The NVIDIA New Jersey office is located at Bell Works in Holmdel, the former Bell Labs location where pioneering deep learning work took place from the mid 80s to the mid 90s and where the first industrial deep learning product was developed, a comme ...Read More
Abstract:
The NVIDIA New Jersey office is located at Bell Works in Holmdel, the former Bell Labs location where pioneering deep learning work took place from the mid 80s to the mid 90s and where the first industrial deep learning product was developed, a commercial handwritten check-reading system. &nbsp; The current work started in the Spring of 2015 with a small core team. By March 2016 the team demonstrated a complete learned driving application on local roads and highways. Since then the team has grown and the capabilities of our on-road driving system have expanded. In addition, an augmented re-simulator has been built that faithfully reproduces real vehicle behavior, allowing to measure progress and system testing before a real vehicle goes on the road. &nbsp; While synthetic simulation can create arbitrary scenarios, programming such scenarios is time consuming and requires careful attention to ensure the simulated sensor data mimics real sensor data closely. Data replay on the other hand uses recorded sensor data but it is limited to open loop simulation. Our augmented re-simulator fills the gap in between by operating from recorded sensor data and applying viewpoint transforms to represent the world for different positions and orientations of the ego car. &nbsp; Our learned system, PilotNet, predicts the paths that a human would follow, in world coordinates, under multiple different driving behaviors such as lane keeping, lane changes, or choosing a fork in the road. Fusing this approach into a traditional system offers important benefits by providing redundant coverage for critical system components, for example, by predicting road geometry when accurate map localization is unavailable.  Back
 
Topics:
Autonomous Vehicles
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9932
Streaming:
Download:
Share:
 
Abstract:
Will the 24hrs of Le Mans redefine the Future of Vehicle Intelligence and Autonomy? After 125 years of racing horseless carriages motorsport is set to fade away in the era of driverless cars. Or is it? The 24hrs of Le Mans has always been a proving g ...Read More
Abstract:
Will the 24hrs of Le Mans redefine the Future of Vehicle Intelligence and Autonomy? After 125 years of racing horseless carriages motorsport is set to fade away in the era of driverless cars. Or is it? The 24hrs of Le Mans has always been a proving ground for innovative automotive technology but is automated driving the only future? Instead, will innovations in artificial Intelligence be used enhance human driver skill and safety, create a future where AI becomes our co-driver &amp; guardian angel. Will AI help us reimagine the human machine interface and enable disabled athletes to compete beyond their physical or neurological limits? Will AI enable races that bridge the gap between the real and virtual worlds? Are we entering the Motorsport Metaverse?  Back
 
Topics:
Autonomous Vehicles
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9933
Streaming:
Download:
Share:
 
Abstract:
We'll discuss how we're working to enable safe, reliable autonomous transport by developing highly accurate, real-time 3D views around autonomous vehicles. Our perception sensor suite uses RADAR, LIDAR, cameras, and IMUs to provide a 360-degree saf ...Read More
Abstract:
We'll discuss how we're working to enable safe, reliable autonomous transport by developing highly accurate, real-time 3D views around autonomous vehicles. Our perception sensor suite uses RADAR, LIDAR, cameras, and IMUs to provide a 360-degree safety shield. We'll explain how data from high-performance imaging RADAR, LIDAR, and cameras are fused together to give the vehicle its sense of sight, while the IMU gives the vehicle its sense of feeling and ensures it maintains its heading. The large amount of data generated by our Drive360 sensors requires in-vehicle high performance AI computers to create a real-time 3D view around the vehicle.  Back
 
Topics:
Autonomous Vehicles, AI Application Deployment and Inference
Type:
Sponsored Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9991
Streaming:
Download:
Share:
Bioinformatics & Genomics
Presentation
Media
Abstract:
We'll discuss the computational challenge of aligning short DNA reads to very large reference genomes, a problem that tests the limits of computing hardware. We'll explain how adapting a CUDA-accelerated short-read aligner to handle these genomes r ...Read More
Abstract:
We'll discuss the computational challenge of aligning short DNA reads to very large reference genomes, a problem that tests the limits of computing hardware. We'll explain how adapting a CUDA-accelerated short-read aligner to handle these genomes resulted in a tenfold reduction in execution time.  Back
 
Topics:
Bioinformatics & Genomics, Performance Optimization
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9350
Streaming:
Download:
Share:
 
Abstract:
We'll discuss how interpretable deep learning can significantly advance our understanding of genomic regulation. All our cells have the same DNA sequence, yet different cell-types express different genes in a process called genomic regulation. This ...Read More
Abstract:
We'll discuss how interpretable deep learning can significantly advance our understanding of genomic regulation. All our cells have the same DNA sequence, yet different cell-types express different genes in a process called genomic regulation. This regulation is driven by binding regulatory proteins to DNA. The vast majority of disease-associated mutations do not disrupt the DNA sequences of genes, but rather disrupt DNA sequences important for regulatory protein binding. Unfortunately, conventional computational models fail to explain which regulatory proteins are impacted for over 90 percent of such mutations. We show that by using deep learning coupled with our interpretation algorithms DeepLIFT (https://github.com/kundajelab/deeplift) and TF-MoDISco (https://github.com/kundajelab/tfmodisco) we can explain a substantially greater fraction of mutations that impact genomic regulation and obtain novel biological insights that are not provided by other methods.  Back
 
Topics:
Bioinformatics & Genomics, Deep Learning and AI Frameworks, Computational Biology and Chemistry
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9632
Streaming:
Download:
Share:
 
Abstract:
We'll describe our work to develop multi-task deep learning models for the improved genetic risk prediction of coronary artery disease. Although studies have shown that basic coronary artery disease genetic-risk prediction models provide modest clin ...Read More
Abstract:
We'll describe our work to develop multi-task deep learning models for the improved genetic risk prediction of coronary artery disease. Although studies have shown that basic coronary artery disease genetic-risk prediction models provide modest clinical utility, improved comprehensive models can make this a reality of clinical practice. This information can help guide therapy decisions and provide an impetus for optimizing lifestyle modifications, thereby improving health outcomes and clinical efficiency. Some preliminary models have been described for autoencoding genetic data, but these models pay no attention to the underlying structure of genetic data. We'll talk about our work to provide best practices for autoencoding genetic data, with the ultimate goal of using these latent genetic factors as input for improved neural network and deep learning-based genetic risk prediction models.  Back
 
Topics:
Bioinformatics & Genomics
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9663
Streaming:
Download:
Share:
 
Abstract:
Learn about the importance of genomics in precision medicine and understand how researchers are decoding genomic information by building deep learning models. We'll show how the Kipoi model zoo for genomics (kipoi.org) can help in this endeavor and ...Read More
Abstract:
Learn about the importance of genomics in precision medicine and understand how researchers are decoding genomic information by building deep learning models. We'll show how the Kipoi model zoo for genomics (kipoi.org) can help in this endeavor and discuss several Kipoi use cases that demonstrate how it facilitates using, sharing, archiving, and building deep learning models in genomics. In addition, we'll highlight some recent successes of deep learning in genomics. Session participants can expect to gain appreciation for sharing end-to-end processing pipelines (not just models) and gain insight into how deep learning and GPU hardware accelerators are changing genomics.  Back
 
Topics:
Bioinformatics & Genomics, AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9705
Streaming:
Download:
Share:
 
Abstract:
Genes fused to one another can drive aggressive cancer cell growth. Fused MYB and NFIB genes are a hallmark of adenoid cystic carcinoma (ACC), but scientists don't yet know how the combined MYB-NFIB protein functions. Learn how we're working to bet ...Read More
Abstract:
Genes fused to one another can drive aggressive cancer cell growth. Fused MYB and NFIB genes are a hallmark of adenoid cystic carcinoma (ACC), but scientists don't yet know how the combined MYB-NFIB protein functions. Learn how we're working to better understand the complex biological interactions that lead to cancer. We'll describe how we investigated the MYB-NFIB fusion protein in ACC by ingesting many large, publicly available biological databases into a colossal hypergraph database designed to preserve the hierarchical structure and relationships inherent in biological data. We'll discuss how we implemented a supervised learning model to identify meaningful patterns that could explain ACC tumor biology. We will also cover how using GPUs results in a thousandfold increase in logistic regression analysis computational efficiency.  Back
 
Topics:
Bioinformatics & Genomics, AI in Healthcare, Accelerated Data Science
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9760
Streaming:
Download:
Share:
 
Abstract:
We will discuss the scientific drivers for moving bioinformatics software to GPU platforms. The DOE Joint Genome Institute generates petabytes of environmental genomics data each year that is shared with thousands of scientists through web-based data ...Read More
Abstract:
We will discuss the scientific drivers for moving bioinformatics software to GPU platforms. The DOE Joint Genome Institute generates petabytes of environmental genomics data each year that is shared with thousands of scientists through web-based data analysis platforms. As sequencing technology changes, the character of the data changes and enables scientists to ask new questions that can be answered with higher fidelity. These data are integrated with the entire corpus of data generated by the DOE JGI, as well as labs around the world, leading to the need for large-scale high-performance computing that leverages the DOE Advanced Scientific Computing Research user facilities. We'll explain how systems like Summit at the Oak Ridge Leadership Computing Facility enable large-scale analysis of multi-omics data, which leads to new discoveries and new hypotheses.  Back
 
Topics:
Bioinformatics & Genomics, Accelerated Data Science
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9945
Streaming:
Download:
Share:
Climate, Weather, Ocean Modeling
Presentation
Media
Abstract:
We'll talk about how we're applying deep learning to weather forecasting at Weather News, one of the world's largest forecasting companies. We're now able to provide Japanese TV news shows with AI-generated weather information, and we plan to exp ...Read More
Abstract:
We'll talk about how we're applying deep learning to weather forecasting at Weather News, one of the world's largest forecasting companies. We're now able to provide Japanese TV news shows with AI-generated weather information, and we plan to expand elsewhere in Asia. We'll explain how we used TensorFlow on an NVIDIA DGX-2 machine and innovative learning model to add measurement results and increase accuracy of our forecaster. We'll also talk about how we're creating new learning models with TensorRT on the DGX-2. We'll touch on other potential uses for our weather technology in settings such as autonomous cars and solar power plants.  Back
 
Topics:
Climate, Weather, Ocean Modeling, Advanced AI Learning Techniques
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9164
Streaming:
Download:
Share:
 
Abstract:
We'll talk about how NOAA and NVIDIA are collaborating to develop deep learning applications to improve use of satellite data in weather forecasting. As model and satellite resolutions increase, data volumes are growing rapidly. AI has the potential ...Read More
Abstract:
We'll talk about how NOAA and NVIDIA are collaborating to develop deep learning applications to improve use of satellite data in weather forecasting. As model and satellite resolutions increase, data volumes are growing rapidly. AI has the potential to help address this issue by automating some aspects of data analysis. We'll discuss projects involving automated detection, translation, enhancement, and emulation of satellite and model data. We've developed networks to automatically detect high-impact weather events and automatically translate satellite observations into model variables. These networks also create high frame rate video from low frame rate satellite loops and generate new physical parameterizations of ground water directly from data. We'll share our results to date and discuss techniques used, lessons learned, and future plans.  Back
 
Topics:
Climate, Weather, Ocean Modeling, Accelerated Data Science
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9217
Streaming:
Download:
Share:
 
Abstract:
The numerical simulations underlying our weather forecasts and climate projections depend on a large set of sub-models called parameterization schemes to represent unresolved processes in the Earth system. Many existing parameterizations are computat ...Read More
Abstract:
The numerical simulations underlying our weather forecasts and climate projections depend on a large set of sub-models called parameterization schemes to represent unresolved processes in the Earth system. Many existing parameterizations are computationally intensive or contain simplifying assumptions that lead to biases or artifacts in the simulations. We'll talk about how machine learning models can address both problems by emulating a more complex parameterization or learning to represent a process from long records of detailed observations. We will describe machine learning parameterizations of the conversion of cloud water to rain and the transfer of energy between the surface and atmosphere, and we'll compare these against existing approaches for these problems.  Back
 
Topics:
Climate, Weather, Ocean Modeling, AI and DL Research
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9245
Streaming:
Download:
Share:
 
Abstract:
We'll discuss the Max Planck/University of Chicago Radiative MHD code (MURaM), the primary model for simulating the sun's upper convection zone, its surface, and the corona. Accelerating MURaM allows physicists to interpret high-resolution solar ob ...Read More
Abstract:
We'll discuss the Max Planck/University of Chicago Radiative MHD code (MURaM), the primary model for simulating the sun's upper convection zone, its surface, and the corona. Accelerating MURaM allows physicists to interpret high-resolution solar observations. We'll describe the programmatic challenges and optimization techniques we employed while using the OpenACC programming model to accelerate MURaM on GPUs and multicore architectures. We will also examine what we learned and how it could be broadly applied on atmospheric applications that demonstrate radiation-transport methods.  Back
 
Topics:
Climate, Weather, Ocean Modeling, Programming Languages
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9288
Streaming:
Download:
Share:
 
Abstract:
Scientific model performance has begun to stagnate over the last decade due to plateauing core speeds, increasing model complexity, and mushrooming data volumes. Learn how our team at the National Center for Atmospheric Research is pursuing an end-to ...Read More
Abstract:
Scientific model performance has begun to stagnate over the last decade due to plateauing core speeds, increasing model complexity, and mushrooming data volumes. Learn how our team at the National Center for Atmospheric Research is pursuing an end-to-end hybrid approach to surmounting these barriers. We'll discuss how combining ML-based emulation with GPU acceleration of numerical models can pave the way toward new scientific modeling capabilities. We'll also detail our approach, which uses machine learning and GPU acceleration to produce what we hope will be a new generation of ultra-fast meteorological and climate models that provide enhanced fidelity with nature and increased value to society.  Back
 
Topics:
Climate, Weather, Ocean Modeling, Accelerated Data Science, HPC and Supercomputing
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9731
Streaming:
Download:
Share:
 
Abstract:
We'll discuss the revolution in computing, modeling, data handling and software development that's needed to advance U.S. weather-prediction capabilities in the exascale computing era. Creating prediction models to cloud-resolving 1 KM-resolution s ...Read More
Abstract:
We'll discuss the revolution in computing, modeling, data handling and software development that's needed to advance U.S. weather-prediction capabilities in the exascale computing era. Creating prediction models to cloud-resolving 1 KM-resolution scales will require an estimated 1,000-10,000 times more computing power, but existing models can't exploit exascale systems with millions of processors. We'll examine how weather-prediction models must be rewritten to incorporate new scientific algorithms, improved software design, and use new technologies such as deep learning to speed model execution, data processing, and information processing. We'll also offer a critical and visionary assessment of key technologies and developments needed to advance U.S. operational weather prediction in the next decade.  Back
 
Topics:
Climate, Weather, Ocean Modeling, AI and DL Research, HPC and AI
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9750
Streaming:
Download:
Share:
Computational Biology and Chemistry
Presentation
Media
Abstract:
Learn how we're bringing Gromacs up to speed with the latest cutting-edge multi-GPU technology. Gromacs, a simulation package for biomolecular systems, is one of the most highly used HPC applications globally. It already benefits from GPU accelerati ...Read More
Abstract:
Learn how we're bringing Gromacs up to speed with the latest cutting-edge multi-GPU technology. Gromacs, a simulation package for biomolecular systems, is one of the most highly used HPC applications globally. It already benefits from GPU acceleration to allow fast simulation of large and complex systems. However, as GPUs become more powerful and increasingly sophisticated multi-GPU systems become available, Gromacs must adapt to optimally benefit from the massive extent of performance on offer. We will describe work to port all significant remaining computational kernels to the GPU, and to perform the required Inter-GPU communications using peer-to-peer memory copies, such that the GPU is exploited throughout and repeated PCIe transfers are avoided. We will present performance results to show the impact of our developments, and also describe the Gromacs performance model we've created to guide our work.  Back
 
Topics:
Computational Biology and Chemistry, Performance Optimization
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9270
Streaming:
Download:
Share:
 
Abstract:
The chemical shift of a protein structure offers a lot of information about the physical properties of the protein. Being able to accurately predict this shift is essential in drug discovery and in some other areas of molecular dynamics research. But ...Read More
Abstract:
The chemical shift of a protein structure offers a lot of information about the physical properties of the protein. Being able to accurately predict this shift is essential in drug discovery and in some other areas of molecular dynamics research. But because chemical shift prediction algorithms are so computationally intensive, no application can predict chemical shift of large protein structures in a realistic amount of time. We explored this problem by taking an algorithm called PPM_One and ported it to NVIDIA V100 GPUs using the directive-based programming model, OpenACC. When testing several different protein structures of datasets ranging from 1M to 11M atoms we observed ~45X average speedup between the datasets and a maximum of a 61X speedup. We'll discuss techniques to overcome programmatic challenges and highlight the scientific advances enabled by the model OpenACC.  Back
 
Topics:
Computational Biology and Chemistry
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9277
Streaming:
Download:
Share:
 
Abstract:
Generative Variational Autoencoders (VAE) in molecular discovery and new materials design have recently gained considerable attention in academia as well as industry (Gomez-Bombarelli, 2017). In this talk, we will present results from a combined Dow ...Read More
Abstract:
Generative Variational Autoencoders (VAE) in molecular discovery and new materials design have recently gained considerable attention in academia as well as industry (Gomez-Bombarelli, 2017). In this talk, we will present results from a combined Dow Chemical and NVIDIA development effort to implement a VAE for chemical discovery. We'll discuss challenges associated with applying deep learning to chemistry and highlight recently developed methods. Highlights from our presentation will include a discussion of methods to analyze and sample from an organized latent representation in a conditioned variational autoencoder, tips for training a complex architecture, distributed multi-node training using Horovod, and results showing the generation of molecular structure with associated property prediction.  Back
 
Topics:
Computational Biology and Chemistry, Advanced AI Learning Techniques
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9417
Streaming:
Download:
Share:
 
Abstract:
We'll discuss cuTENSOR, a high-performance CUDA library for tensor operations that efficiently handles the ubiquitous presence of high-dimensional arrays (i.e., tensors) in today's HPC and DL workloads. This library supports highly efficient tensor ...Read More
Abstract:
We'll discuss cuTENSOR, a high-performance CUDA library for tensor operations that efficiently handles the ubiquitous presence of high-dimensional arrays (i.e., tensors) in today's HPC and DL workloads. This library supports highly efficient tensor operations such as tensor contractions (a generalization of matrix-matrix multiplications), point-wise tensor operations such as tensor permutations, and tensor decompositions (a generalization of matrix decompositions). While providing high performance, cuTENSOR also allows users to express their mathematical equations for tensors in a straightforward way that hides the complexity of dealing with these high-dimensional objects behind an easy-to-use API. CUDA 10.1 enables CUDA programmers to utilize Tensor Cores directly with the new mma.sync instruction. In this presentation, we describe the functionality of mma.sync and present strategies for implementing efficient matrix multiply computations in CUDA that maximize performance on NVIDIA Volta GPUs. We then describe how CUTLASS 1.3 provides reusable components embodying these strategies. CUTLASS 1.3 demonstrates a median 44% speedup of CUDA kernels executing layers from real-world Deep Learning workloads.  Back
 
Topics:
Computational Biology and Chemistry, Tools and Libraries, HPC and AI
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9593
Streaming:
Download:
Share:
 
Abstract:
We'll discuss advanced implementations of molecular dynamics for studying the motions of biochemical systems and dissecting free energies in drug binding and molecular recognition. Attendees should have basic knowledge of CUDA. We'll focus on strat ...Read More
Abstract:
We'll discuss advanced implementations of molecular dynamics for studying the motions of biochemical systems and dissecting free energies in drug binding and molecular recognition. Attendees should have basic knowledge of CUDA. We'll focus on strategies for parallelism based on the computer science of how graphics cards operate. We'll also talk about the results for applications in pharmaceutical and academic computational biology.  Back
 
Topics:
Computational Biology and Chemistry
Type:
Talk
Event:
GTC Silicon Valley
Year:
2019
Session ID:
S9658
Streaming: