No
Yes
View More
View Less
Working...
Close
OK
Cancel
Confirm
System Message
Delete
My Schedule
An unknown error has occurred and your request could not be completed. Please contact support.
Scheduled
Wait Listed
Personal Calendar
Speaking
Conference Event
Meeting
Interest
Coming Soon
Conflict Found
This session is already scheduled at another time. Would you like to...
Loading...
Please enter a maximum of {0} characters.
{0} remaining of {1} character maximum.
Please enter a maximum of {0} words.
{0} remaining of {1} word maximum.
must be 50 characters or less.
must be 40 characters or less.
Session Summary
We were unable to load the map image.
This has not yet been assigned to a map.
Search Catalog
Reply
Replies ()
Search
New Post
Microblog
Microblog Thread
Post Reply
Post
Your session timed out.
This web page is not optimized for viewing on a mobile device. Visit this site in a desktop browser to access the full set of features.
2017 GTC Washington DC
Add to My Interests
Remove from My Interests
 

DC7243 - Accelerated Deep Learning Advances in HPC Recent advances in the deployment of deep learning recurrent nets have been demonstrated in scaling studies of Princeton's new Deep Learning Code -- "FRNN (Fusion Recurrent Neural Net) Code on modern GPU systems. This is a "big-data" project in that it has access to the huge EUROFUSION/JET disruption data base of over a half-petabyte to drive these studies. FRNN implements a distributed data parallel synchronous stochastic gradient approach with TensorFlow and Theano libraries at the backend and MPI for communication. This deep learning software has recently demonstrated excellent scaling up to 6,000 GPUs on Titan at Oak Ridge National Lab. The associated accomplishments exhibit clear progress toward the goal of establishing the practical feasibility of using leadership-class supercomputers to greatly enhance training of neural nets for transformational impact on key discovery science application domains such as fusion energy science. Powerful systems expected to be engaged for near-future deployment of this deep learning software include: (1) NVIDIA's SATURN V – featuring its nearly 1,000 Pascal P100 GPUs; (2) Switzerland's Piz Daint Cray XC50 system with 4,500 P100 GPUs; (3) Japan's Tsubame 3 system with 3,000 P100 GPUs; (4) and OLCF's Summit-Dev system. Summarily, deep learning software trained on large scientific datasets hold exciting promise for delivering much-needed predictive tools capable of accelerating knowledge discovery. The associated creative methods being developed – including a new half-precision capability -- also has significant potential for cross-cutting benefit to a number of important application areas in science and industry. 25 Minute Talk William Tang - Professor, Princeton University
Add to My Interests
DC7111 - Accelerating Cyber Threat Detection with GPUs Analyzing vast amounts of enterprise cyber security data to find threats can be cumbersome. Cyber threat detection is also a continuous task, and because of financial pressure, companies have to find optimized solutions for this volume of data. We'll discuss the evolution of big data architectures used for cyber defense and how GPUs are allowing enterprises to efficiently improve threat detection. We'll discuss (1) briefly the evolution of traditional platforms to lambda architectures and ultimately GPU-accelerated solutions; (2) current GPU-accelerated database, analysis tools, and visualization technologies (such as MapD, BlazingDB, H2O.ai, Anaconda and Graphistry), and discuss the problems they solve; (3) the need to move beyond traditional rule based indicators of compromise and use a combination of machine learning, graph analytics, and deep learning to improve threat detection; and finally (4) our future plans to continue to advance GPU accelerated cyber security R&D as well as the GPU Open Analytics Initiative. 50 Minute Talk Josh Patterson - Director of Applied Solutions Engineering, NVIDIA
Add to My Interests
DC7249 - Accelerating Your VR Applications with the VRWorks SDK Across graphics, audio, video, and physics, the NVIDIA VRWorks suite of technologies helps developers maximize performance and immersion for VR applications. We'll explore the latest features of VRWorks, explain the VR-specific challenges they address, and provide application-level tips and tricks to take full advantage of these features. Special focus will be given to the latest VRWorks integrations into Unreal Engine and Unity. 25 Minute Talk Victoria Rege - Global Alliances & Ecosystem Development, VR , NVIDIA
Add to My Interests
DC7255 - Activating Tissue Data in the Era of Computational Medicine Pathology departments and translational research centers have amassed invaluable, untapped information in the form of glass pathology slides. The recent proliferation of digital pathology has shifted pathology into the era of computational medicine by creating the opportunity to quantify and integrate tissue data to supplement the existing cancer model centered around human expertise, and corresponding patient history and "-omic" data. With the help of clinical partners and massive computing infrastructures provided by NVIDIA, we are developing deep learning powered tools that activate those digital slides to address problems in the clinic and inform disease prognosis and therapeutic plans. This talk will outline our approach to deep learning based pathology image processing, including coarse feature embedding using fully convolutional networks and stain normalization using Generative Adversarial networks. It will also highlight a few of our recent successes in this domain including image classification applications for identifying metastases in breast and gastric lymph nodes with and image biomarker generation using deep learning to predict lymph node metastasis from the primary tumor histology. 50 Minute Talk Hunter Jackson - Chief Scientific Officer, Proscia Inc
Add to My Interests
DC7200 - Adapting Deep Learning to New Data Using ORNL's Titan Supercomputer There has been a surge of success in using deep learning as it has provided a new state of the art for a variety of domains. While these models learn their parameters through data-driven methods, model selection through hyper-parameter choices remains a tedious and highly intuition-driven task. We've developed two approaches to address this problem. Multi-node evolutionary neural networks for deep learning (MENNDL) is an evolutionary approach to performing this search. MENNDL is capable of evolving not only the numeric hyper-parameters, but is also capable of evolving the arrangement of layers within the network. The second approach is implemented using Apache Spark at scale on Titan. The technique we present is an improvement over hyper-parameter sweeps because we don't require assumptions about independence of parameters and is more computationally feasible than grid-search. 25 Minute Talk Travis Johnston - Postdoctoral Research Associate, Oak Ridge National Laboratory
Steven Young - Research Scientist in Deep Learning, Oak Ridge National Laboratory
Add to My Interests
DC7139 - Advanced Analytics and Machine Learning with Geospatial Data: A World of Possibilities As data sources such as sensors, social media, mobile, cloud, and machine logs become pervasive within the enterprise and the 4Vs of data managed by modern business applications increase exponentially, organizations need faster and better visualizations to explore data at scale and discover game-changing insights. Organizations are increasingly adopting modern technologies such as GPUs to interactively visualize billions of data elements in real-time and take faster business decisions. You'll learn how you can augment your existing business applications and visualization capabilities with GPU-accelerated rendering of maps and accompanying dashboards for spatial awareness and location-based analytics, and how GPU's massive parallelization minimizes the need for data preparation and sampling and delivers brute force capabilities to interactively visualize the fast-moving location and time-series data such as trading, traffic, social media, and vehicle telematics at scale and with speed. 25 Minute Talk Nohyun Myung - Principal Architect for Location-Based Analytics, Kinetica
Add to My Interests
DC7191 - Aerial AI: How Drones and Deep Learning are Building the Cities of the Future Learn how drones and artificial intelligence are making construction of sites smarter, safer, and more efficient. As jobsites scale and become increasingly complex with more machines, people, and assets, data is impossible to capture and industrial site managers lose oversight and control. Drones and AI systems give us the power to capture, store, and analyze data about activity occurring at every moment on site. We'll discuss deep learning algorithms and neural network architectures that enable us to analyze data and break down images and videos into an ontology of objects, actions, relationships, and patterns that ultimately provides visibility into jobsites. In addition, we'll dive into edge computing and how to deploy an AI platform across thousands of edge nodes that use high-powered GPUs to capture, transfer, and process data, even in remote sites without internet connectivity. 25 Minute Talk Angela Sy - Director, AI & Strategy, Skycatch
Add to My Interests
DC7129 - AI at the Edge - Intelligent Machines Artificial intelligence is impacting almost every part of the industrial and agricultural supply chain. From robots that quickly adapt to build new products, to automated vehicles that address last-mile challenges for product delivery, to UAVs that can automatically detect failing infrastructure, the world is transitioning from processes that are largely manual to ones that are largely automated. We'll discuss how AI and deep learning are enabling these advances. We'll also analyze a sampling of early successes across different applications. And finally we'll describe some of the remaining challenges to wide-scale deployment, and the work NVIDIA is doing to address those challenges via its Isaac initiative. 25 Minute Talk Jesse Clayton - Senior Manager of Product Management for Intelligent Machines, NVIDIA
Add to My Interests
DC7177 - AI, HPC Infrastructure, & Visualizations on Microsoft Azure Government Learn how Microsoft Azure Government enables compliant deep learning and traditional HPC-based workloads using powerful NVIDIA Tesla GPU accelerators and scale out using Azure's low-latency networking backed by InfiniBand infrastructure. Find out about the roadmap to support visualization scenarios and hear about customer stories and partner solutions that are leveraging these platform capabilities. 50 Minute Talk Kyle Deeds - Capacity and Specialized Hardware Lead, Azure Government Engineering, Microsoft
Zach Kramer - Engineer Lead, Azure Government, Microsoft
Add to My Interests
DC7246 - AI in Healthcare: Beyond the Hype Cycle AI is transforming healthcare" is the buzz around every news alert these days—but is it true? Where is AI and deep learning affecting healthcare and how is it impacting the medical imaging space? Join a thought leadership panel as government, industry and academic experts discuss the real calibration of the space—separating reality from the noise to explore how deep learning is advancing clinical practice, including advancements to overcome data and regulatory challenges. 50 Minute Panel Mike Tilkin - Chief Information Officer & Executive Vice President, American College of Radiology
Carla Leibowitz - Head of Strategy and Marketing, Arterys
Abdul Hamid Halabi - Global Business Development Lead, Healthcare & Life Sciences , NVIDIA
Katherine P. Andriole - Associate Professor of Radiology, Harvard Medical School, MGH & BWH Center for Clinical Data Science
Ronald Summers - Senior Investigator, NIH Clinical Center
Elliot Fishman - Professor of Radiology and Oncology , Johns Hopkins Hospital
Abdul Hamid Halabi - Global Business Development Lead, Healthcare & Life Sciences , NVIDIA
Add to My Interests
DC7114 - AI: Transforming Your Work and the World Now Artificial intelligence is changing the world at an accelerating pace. AI is now rapidly moving from its academic roots to revolutionize business and industry through a convergence of technology advances, social transformations, and genuine economic needs. In this keynote, Greg will describe how AI will improve and create new jobs as it transforms every corner of the modern economy -- from energy, transportation, and healthcare to manufacturing, entertainment, education, IT, and government. 60 Minute Keynote Greg Estes - Vice President Developer Programs, NVIDIA
Add to My Interests
DC7260 - Alexa, Tell me how Kaldi and Deep Learning Revolutionized Automatic Speech Recognition We'll review the history of automatic speech recognition (ASR) technology, and show how deep neural networks have revolutionized the field within the last 5 years, giving birth to Alexa, enhancing Siri and nudging Google Home to market, and generally making ASR a household phenomenon. The story will be told from the viewpoint of Kaldi , a widely used set of open-source ASR tools in both academia and industry, touching on some key milestones and seminal developments along this short-yet-exciting journey, such as suitable network architectures, novel training criteria, and scalable optimization algorithms, along with prescient research funding, realistic data sets, and competitive benchmark tests conducted by neutral entities. 50 Minute Talk Sanjeev Khudanpur - Associate Professor of Electrical and Computer Engineering, The Johns Hopkins University
Add to My Interests
DC7157 - A Look Inside the US Navy's Mixed Reality Lab Join us for a look inside the U.S. Navy's mixed reality lab, including an overview of Unity-based AR/VR projects to support the warfighter. We'll highlight both the recent live-fire exercise aboard the U.S.S. Bunker Hill, performed to test an augmented reality heads-up display as part of a topside ship gunnery system, and the Secretary of the Navy Innovation Award-winning work the team is doing to create 3D lidar scans of the entire U.S. Navy fleet and a collaboration environment that supports ship installations. 50 Minute Talk Joshua Harguess - Research Scientist, Gunnery System with Augmented Reality (GunnAR) Lea, SPAWAR Systems Center, Pacific
Add to My Interests
DC7216 - A Machine Learning Approach for Tissue Biomechanics We're developing a machine learning strategy for improving lung and liver cancer treatments. Specifically, near-real-time lung and liver tissue elasticity estimations using only deformation maps that would typically be available from 4DCT lung images. The key technical innovation presented by our work is the integration of patient-specific, GPU-based biomechanical models with GPU-based machine learning approaches. For learning purposes, we employed TensorFlow. Enabling such measurements within the radiotherapy setup facilitates multiple advancements, including: (a) functional lung and liver preserving radiotherapy treatment planning, where the patient's treatment efficacy and quality of life can be significantly improved, (b) patient treatment response monitoring, where the impact of treatment on the patient can be monitored, and (c) novel therapeutic approaches, where the treatment can be adapted to the patient's disease conditions. 25 Minute Talk Anand Santhanam - Associate Professor, University of California, Los Angeles.
Add to My Interests
DC7227 - An Overview of AI on Amazon Web Services Amazon Web Services offers a family of intelligent services that provide cloud-native machine learning and deep learning technologies to address different use cases and needs. For developers looking to add managed AI services to their applications, AWS brings natural language understanding and automatic speech recognition with Amazon Lex, visual search and image recognition with Amazon Rekognition, text-to-speech with Amazon Polly, and developer-focused machine learning with Amazon Machine Learning. The AWS Deep Learning Amazon Machine Images provides a way for AI developers and researchers to quickly and easily begin using any of the major deep learning frameworks (Apache MXNet, TensorFlow, Caffe, and others) to train sophisticated, custom AI models; experiment with new algorithms; and learn new DL skills and techniques on AWS' massive GPU-accelerated compute infrastructure. We'll provide a meaningful overview of how to improve scale and efficiency of AI applications on the AWS Cloud. 50 Minute Talk Vikram Madan - Senior Product Manager, AWS Deep Learning, Amazon Web Services
Add to My Interests
DC7258 - Applying Neural Networks to Enhance Drug Discovery Berkeley Lights has developed unique capabilities to select, sort and annotate single live cells using light with a nanofluidic chip.  By combining this capability with deep learning, Convolutional Neural Networks and Nvidia GPUs, BLI can accurately identify, characterize and assay cells.  Applying this to antibody discovery for novel therapeutics, the Beacon platform automatically identifies the rare cells out of thousands that are producing the antibody of interest. This can reduce timelines from months to less than a week allowing scientist to iterate significantly faster. 25 Minute Talk John Tanney - Director of Algorithms, Berkeley Lights
Add to My Interests
DC7171 - Applying Virtual Reality, and Augmented Reality to the Lifecycle Phases of Complex Products We'll demonstrate how to best utilize GPU technology for virtual reality and augmented reality applications. The lifecycle phases of a product such as a naval ship include design, production, deployment, operations, maintenance, and upgrades. Early in the design phase, it's typical to have the product modeled in a 3D CAD system. By adding physical properties to the CAD model, a virtual model can be created and used to perform physics-based simulations. These simulations are used to verify and validate the design. This virtual product model can be presented and manipulated in VR. The goal is to catch problems during the conceptual design phase before construction begins. VR and AR can both be applied during construction, operations, maintenance, and product upgrades. The Lockheed Martin Surface Navy Innovation Center (LM-SNIC) is investigating the best practices for applying VR and AR to the lifecycle phases of complex products, such as naval ships, aircraft, and ground radars. 50 Minute Talk Christopher Crouch - Associated Member of Engineering Staff, Lockheed Martin
Rich Rabbitz - Principal Member of Engineering Staff, Lockheed Martin
Add to My Interests
DC7185 - Artificial Intelligence and Future of Genomic Medicine Each of us shares a common genome, which serves as a blueprint for our health and biology, that is 99.9 percent identical among individuals. Understanding the impact of the 0.1 percent of genomic differences among individuals is crucial for understanding the causes of diseases and developing personalized treatments. Genome interpretation represents an area where humans have little innate capability, and clinicians will need to rely on artificial intelligence to guide their decisions. A significant challenge is to move beyond black box AI, providing insights from the genome in a way that complements human health care providers. In addition, current AI techniques are largely optimized for vision and speech. Novel artificial intelligence algorithms and architectures will be required to optimize the interpretation of genomic data and realize the dream of personalized medicine. 25 Minute Talk Kyle Farh - Director, Data Sciences, Illumina
Add to My Interests
DC7161 - A Sparse Dynamic Graph and Matrix Data Structure Sparse data computations are ubiquitous in science and engineering. Two widely used applications requiring sparse data computations are graph algorithms and linear algebra operations such as sparse matrix-vector multiplication (SpMV). In contrast to their dense data counterparts, sparse-data computations have less locality and more irregularity in their execution -- making them significantly more challenging to optimize. While there are several existing formats for sparse data representations, most of these are restricted to static datasets. We'll show a new data structure called Hornet for dynamic graphs and matrices that scales to extremely large graphs. Hornet can be used for both static and dynamic graph-based problems. 25 Minute Talk Oded Green - Research Scientist, Georgia Tech
Add to My Interests
DC7236 - Augmented & Virtual Reality in Education Educational institutions are driving cutting-edge AR and VR research across many different industries. AR and VR have the potential to change the way students and doctors learn, how designers and engineers create products, and how games and films are made. We'll discuss how the latest research and work at universities can be applied to government and enterprises. The panel will also examine how AR and VR are being used to solve a number of different challenges, including graphics, positional tracking, and more. 50 Minute Panel Nicholas Jushchyshyn - Program Director, Drexel University Animation Capture & Effects Lab
Victoria Rege - Global Alliances & Ecosystem Development, VR , NVIDIA
Ryan McKindles - Technical Lead, MIT
Matthias Zwicker - Professor, Dept. of Computer Science, University of Maryland
Vitalya Berezina-Blackburn - Animation and Motion Capture Specialist, Advanced Computing Center for the Arts and Design at the Ohio State University
Nicholas Jushchyshyn - Program Director, Animation, Visual Effects & Immersive Media, Drexel University
Victoria Rege - Global Alliances & Ecosystem Development, VR , NVIDIA
Add to My Interests
DC7170 - Automated Cotton Weeding We'll present how our see-and-spray machines are being deployed in cotton fields across the southern U.S. to tackle a problem farmers are fighting every day: weed pressure. 25 Minute Talk Lee Redden - CTO, Blue River Technology
Add to My Interests
DC7143 - Autonomous Capabilities of the Joint Tactical Aerial Resupply Vehicle We'll discuss the latest updates and features to the autonomous precision landing and GPS-denied navigation capabilities of the Joint Tactical Aerial Resupply Vehicle (JTARV) platform. These capabilities are enabled by our high-performance computer vision libraries, Sentinel and HawkEye, both of which capitalize on NVIDIA's mobile GPUs and optimized deep learning frameworks. Autonomous navigation for aerial vehicles demand that core algorithms provide not only relevant, actionable information, but that they do so in a timely manner -- that is, the algorithms must operate in real time. We'll discuss how Sentinel object detection networks limit processing requirements for the autonomous precision landing capability. The requirement for high performance dictates optimization at every level, which is the focus of our ongoing research and development efforts. 25 Minute Talk Shawn Recker - Research Scientist, Survice
Add to My Interests
DC7218 - Bringing DeepLearni.ng to the Enterprise: Applications of AI in Finance and Beyond Data-rich enterprise environments are gearing up for deep learning and artificial intelligence: these profound tools will transform how we tackle business opportunities or even revolutionize industries. But the journey of introducing AI to the enterprise is far from simple and is fraught with many challenges and unknowns. We'll systematically dissect technical and non-technical obstacles commonly faced when implementing and deploying AI at the enterprise. Presenting applications in finance and beyond, we'll discuss DeepLearni.ng's approach to overcome these obstacles in a comprehensive, results-driven manner. 25 Minute Talk Eric Kin-Ho Lee - Co-Founder and Co-CEO, DeepLearni.ng
Add to My Interests
DC7122 - Building the Foundation of America's AI-Enabled Cities Why does building AI Cities matter now for America? Why should the U.S. industry and government aggressively develop and deploy AI and deep learning to solve important problems around public safety and operational efficiency in our urban centers? What are the global trends that make this the right time to drive these changes? We'll cover these topics and more. 25 Minute Talk Adam Scraba - Global Business Development Lead, NVIDIA
Add to My Interests
DC7235 - Caffe2: A New Lightweight, Modular, and Scalable Deep Learning Framework Caffe2 is a lightweight, modular, and scalable deep learning framework refactored from the previous Caffe. Caffe2 has been widely used at Facebook to enable new AI & AR experiences. We'll explain some framework basics, the strengths of Caffe2, large scale training support and will walk you through several product use-cases at Facebook including computer vision, machine translation, speech recognition and content ranking. We will also talk about how users benefit from Caffe2's built-in neural network model compression, fast convolution for mobile CPUs, and GPU acceleration. Also we will talk about the integration between ONNX and Caffe2. Open Neural Network Exchange (ONNX) is the first step toward an open ecosystem that empowers AI developers to choose the right tools as their project evolves. ONNX provides an open source format for AI models. It defines an extensible computation graph model, as well as definitions of built-in operators and standard data types. 50 Minute Talk Bram Wasti - Software Engineer, Facebook
Add to My Interests
DC7251 - Classifying Structured Data with TensorFlow Estimators Estimators offer a high-level API with familiar methods such as train, evaluate, and predict. We'll introduce estimators and demonstrate how you can use them to train a deep neural network to classify structured data at scale. We'll also introduce feature engineering techniques, like bucketing, crossing, and embedding. 25 Minute Talk Josh Gordon - Developer Advocate, Google
Add to My Interests
DC7140 - Codesigning Cognitive Computing Systems and Applications The development of cognitive computing applications is at a critical juncture with tough challenges but ample opportunities for great breakthroughs. Many of the cognitive solutions involved are complex and methods required to develop them remain poorly understood. Any major breakthrough in improving such understanding would require large-scale experimentation and extensive data-driven development. In short, we are witnessing the formation of a new modality of programming and even a new modality of application execution. The IBM-Illinois Center for Cognitive Computing Systems Research (C3SR) is developing scalable cognitive solutions that embody both advanced cognitive computing workloads and optimized heterogeneous computing systems for these cognitive workloads. The two streams of research not only complement, but also empower each other, and thus should be carried out in a tightly integrated fashion. 50 Minute Talk Wen-mei Hwu - Professor and Sanders-AMD Chair, University of Illinois at Urbana-Champaign
Add to My Interests
DC7254 - Computational Pathology in Practice: From Cluster to Clinic Pathology is in the midst of a revolution from a qualitative to a quantitative discipline. This transformation is being fueled by machine and deep -learning. At Memorial Sloan Kettering we're building a computational pathology AI using an NVIDIA GPU cluster and a petabyte of clinical data. The goal is to transition the microscopic histological assessment of tissues from a manual and subjective process to a quantitative one. An AI that enables pathologists to efficiently and with speed, perform a pathological assessment that is more reproducible and objective. An AI that will facilitate large-scale, quantitative screening for correlations between tissue morphology and genetic panels like MSK-IMPACT. 25 Minute Talk Thomas Fuchs - Associate Professor, Memorial Sloan Kettering Cancer Center
Add to My Interests
DC7115 - Connection Between Basic Research, AI and US Leadership Our nation's basic research programs, which are primarily funded by the government, have for decades produced amazing results for our country. Countless companies can point to fundamental technologies resulting from basic research they utilize that are the basis for their products or are the essential ingredient. Are we investing enough in basic research? What role should the U.S. government play in these areas? What is the promise of AI that needs more basic R&D to fulfill and how should we do it? 50 Minute Panel France Córdova - Director, National Science Foundation
Robie Samanta Roy - VP of Technology Strategy and Innovation, Lockheed Martin
Daniel Larson - Professor of Physics, Penn State University
William Vanderlinde - Chief Scientist, IARPA
David Luekbe - Vice President of Graphics Research, NVIDIA
Add to My Interests
DC7153 - Convergence Among Disciplines and Technologies: AI, Science and Engineering Research, and HPC AI and machine learning techniques are finding increasing application across all areas of science and engineering, with high performance computing, data and networking playing a critical role in enabling research advances in these areas. In this talk we overview current and planned investments by the National Science in such advanced research cyberinfrastructure. We illustrate the use and promise of these technologies drawing on NSF-funded research at the intersection of a domain science, AI, and advanced research cyberinfrastructure. We'll also discuss how these investments are well aligned with NSF's forward-looking "Big Ideas." 50 Minute Talk James Kurose - Assistant Director, National Science Foundation
Add to My Interests
DC7184 - Count-Ception: Counting by Fully Convolutional Redundant Counting Counting objects in digital images is a process that should be replaced by machines. This tedious task is both time-consuming and prone to errors due to fatigue of human annotators. The goal is to have a system that takes an image as input and returns a count of the objects inside and justification for the prediction in the form of object localization. We repose a problem, originally posed by Lempitsky and Zisserman, to instead predict a count map, which contains redundant counts based on the receptive field of a smaller regression network. We'll discuss a state-of-the-art method for counting objects in images and demonstrate its effectiveness on transmitted light microscopy images. 25 Minute Talk Joseph Paul Cohen - Postdoctoral Fellow , Montreal Institute for Learning Algorithms
Add to My Interests
DC7214 - Counting Passenger Vehicles Using Satellite Imagery Learn the challenges surmounted in developing a vehicle counting algorithm from 30-centimeter satellite imagery. Counting cars from 30-cm resolution imagery is a much more challenging problem than counting cars from even higher resolution aerial (drone) imagery with only a few centimeters. And, since vehicles are composed of both metal and glass, and since light easily passes through glass, the vehicles displayed in satellite imagery are disjointed pixel blobs (that is, a mixture of dark and bright pixels). We'll discuss the challenges faced in developing a detection and counting deep learning algorithm that leverages NVIDIA's GPU capabilities, and how we arrived at a viable solution that is composed of fusing two convolutional neural networks for initially performing segmentation for vehicle detection followed by a more refined localization of the vehicles within each detected vehicle segment using morphological operations, non-maximum suppression, and connected components. 25 Minute Talk Kevin Green - Machine Learning and Computer Vision Scientist, DigitalGlobe Inc.
Mahmoud Lababidi - Senior Data Scientist, DigitalGlobe Inc.
Add to My Interests
DC7127 - Cross-Domain Face Recognition Solution Based On GPU-Powered Deep Learning and Inference Government agencies and commercial companies demonstrate high demand for versatile, stable, and highly efficient person identification solutions supporting cross-domain face recognition and person database clusterization in both controlled and uncontrolled scenarios. Now it's possible to resolve cross-domain face recognition challenges using deep learning and even tasks of quadratic complexity using GPU-powered inference of CNN-based face recognition algorithms. We'll focus on (1) the concept of the GPU-powered platform for cross-domain face recognition; (2) its essential performance and critical technical characteristics; (3) an approach to reaching the demanded efficiency and quality by using the NVIDIA GPU; and (4) providing examples of completed and ongoing projects that demonstrate achieved high-performance and quality parameters in real-life conditions. 25 Minute Talk Alexander Khanin - CEO, VisionLabs
Add to My Interests
DC7146 - CUDA 9 and Beyond CUDA is NVIDIA's parallel computing platform and programming model. In this talk, you'll learn about the new features in CUDA 9, including Cooperative Groups, an extension to the CUDA programming model for organizing groups of threads that helps you write efficient parallel algorithms that are safe, modular, and maintainable. You'll also learn about new features in CUDA that allow you to program the new Volta GPU architecture features such as Tensor Cores, providing the highest possible performance for deep learning training and inference. Finally, you'll get a preview of features and improvements coming to future releases of CUDA, and gain insight into the philosophy driving the development of CUDA and how it will take advantage of current and future GPUs. 50 Minute Talk Mark Harris - Chief Technologist, NVIDIA
Add to My Interests
DC7112 - CUDA Optimization Tips, Tricks and Techniques Optimizing your code can be one of the most challenging tasks in GPU programming, but also one of the most rewarding: the performance difference between an initial version and well-tuned code can be a factor of 10 or more. Some optimizations can be quite straightforward while others require care and deep understanding of how the code is executing. A particular focus will be on optimization of the CPU part of your code, which is frequently overlooked even though it is often easier to tune and just as effective. Sometimes the biggest obstacle is just knowing what to look for, so we'll cover a range of techniques that everyone from beginners to CUDA ninjas might not have thought of before. 50 Minute Talk Stephen Jones - Principal Software Engineer, NVIDIA
Add to My Interests
DC7128 - Current State of Autonomous Video Security at Enterprise Scale We'll explore how UMBO CV is leveraging deep learning techniques and GPUs to scale up how video is analyzed in real time and at enterprise scale. 50 Minute Talk Shawn Guan - CEO, Umbo CV Inc.
Add to My Interests
DC7110 - Deep Learning Demystified What is deep learning? In what fields is it useful, and how does it relate to artificial intelligence? Join this session to get a working understanding of deep learning and why this powerful new technology is getting so much attention. Learn how deep neural networks are trained to perform tasks with superhuman accuracy, and the challenges organizations face in adopting this new approach. We'll also cover the software, hardware, and training resources that many organizations are using to overcome the challenges and deliver breakthrough results. Finally, we'll review several best practices for your first deep learning project. 50 Minute Talk Will Ramey - Director, Developer Programs, NVIDIA
Add to My Interests
DC7172 - Deep Learning Deployment with NVIDIA TensorRT We'll explore some of the common challenges with deep learning deployment and how they can be addressed with NVIDIA TensorRT. Through an example, we'll review a typical workflow for taking a trained deep neural network to production to achieve desired throughput, latency, and energy-efficiency requirements. 50 Minute Talk Shashank Prasanna - Product Marketing Manager, NVIDIA
Add to My Interests
DCL7102 - Deep Learning for Genomics using DragoNN with Keras and Theano In this lab, we use the dragonn toolkit on simulated and real regulatory genomic data, demystify popular DragoNN (Deep RegulAtory GenOmics Neural Network) architectures and provide guidelines for modeling and interpreting regulatory sequence using DragoNN models. We will answer questions such as: When is a DragoNN good choice for a learning problem in genomics? How does one design a high-performance model? And, more importantly, can we interpret these models to discover predictive genome sequence patterns to gain new biological insights? 120 Minutes Instructor-Led Lab Hoo Chang Shin - Certified Instructor, NVIDIA
Add to My Interests
DC7186 - Deep Learning for Real Time Inventory Tracking We'll discuss how IFM uses deep learning for inventory tracking in logistics and manufacturing. 25 Minute Talk Marc Gyongyosi - CEO, Intelligent Flying Machines, Inc.
Add to My Interests
DC7123 - Deep Learning for Real-time Threat Detection - From Active Shooters to Armed Robbery We'll explore how deep learning techniques can be used to transform passive surveillance systems into active threat-detection platforms for environments that range from retail, cities, and campuses. Deep Science is deploying deep learning solutions to spot robberies and assaults as they're occurring in real time. 25 Minute Talk Sean Huver - Founder and CEO, Deep Science
Add to My Interests
DC7126 - Deep Learning in Medical Imaging-Applications in Ophthalmology, Radiology and Oncology Deep learning, facilitated by advances in hardware, software, and algorithms, has emerged as a leading technology in computer vision and image analysis and is being applied to medical imaging with early successes in radiology, oncology, ophthalmology, pathology, and others domains. We'll share our experiences addressing clinical questions in ophthalmology and radiology. In ophthalmology, we have created an automated system for classifying images for retinopathy of prematurity (ROP) using deep learning, while in radiology, we have developed tools for the cancer diagnosis and assessment of response to therapy. 50 Minute Talk Jayashree Kalpathy-Cramer - Director, QTIM lab, MGH/Harvard Medical School
Add to My Interests
DC7137 - Deep Learning in Pulmonary Image Analysis with Incomplete Training Samples This session discusses the topic of computerized pulmonary image analysis from CT scans: the tasks, the unique challenges, and the solutions under deep learning framework. Specifically, we first provide the background of the problem, and limitations of conventional methods. Then, we list the major differences between natural and medical images, especially for pulmonary image analysis. Multiple common tasks will be covered including pathological lung segmentation, lobe fissure estimation, airway extraction, and disease pattern classification. For each candidate task, we then present its distinct features, which lead to a specific choice of deep learning structures. In addition, we discuss how to generate the training data that is usually constraint by limited clinical resources, and assess the design, applicability, performance, and generalizability of deep learning based methods. 25 Minute Talk Ziyue Xu - Staff Scientist, National Institutes of Health
Add to My Interests
DC7151 - Deep Learning with MATLAB: From Concept to Embedded Code Learn how to design a deep learning algorithm in MATLAB and deploy to an embedded Tegra platform, including Jetson TK1, TX1, TX2, and DRIVE PX boards. The workflow starts with algorithm design in MATLAB, which enjoys universal appeal among engineers and scientists because of its expressive power and ease of use. Algorithms used include deep learning augmented with traditional computer vision. Then, networks are trained using NVIDIA GPUs and parallel computing support in MATLAB either on the desktop, a local compute cluster, or in the cloud. Finally, a compiler auto-generates portable and optimized CUDA code from the MATLAB algorithm, which is then cross-compiled and deployed to the Tegra board. Generated code is highly optimized and we present benchmarks that show that performance of generated code is about two-and-a-half times faster than mxNet, about five times faster than Caffe2; about seven times faster than TensorFlow; and is on par with an optimized TensorRT implementation. 50 Minute Talk Ram Cherukuri - Product Marketing, MathWorks
Avinash Nehemiah - Product Manager- Computer Vision and Deep Learning , MathWorks
Add to My Interests
DC7198 - DeepStream : Towards Large-scale Deployment of Intelligent Video Analytics Systems We'll introduce DeepStream, NVIDIA's solution for high-performance video analytics. One of the grand challenges of AI is to understand video content. Applications are endless: video surveillance, live video streaming, ad placement, and more. The problem is that deep learning, which has been boosting modern AI, is computationally expensive. It's even more challenging when it comes to live stream video. That's why we're building the NVIDIA DeepStream SDK, which simplifies development of high-performance video analytics applications powered by deep learning. It's built on top of the NVIDIA Video SDK, which can leverage the GPU's hardware encoding and decoding horsepower, and on top of NVIDIA TensorRT, which accelerates deep neural network's inferencing. We have seen successful large-scale deployment of such intelligent video analytics systems, and we see this as an unstoppable trend. 50 Minute Talk Jeremy Furtek - Senior Developer Technology Engineer, NVIDIA
Add to My Interests
DC7144 - DGX System Deep-Dive: Best Practices for Scalable Deep Learning Performance NVIDIA DGX Systems powered by Volta deliver breakthrough performance for today's most popular deep learning frameworks. Attend this session and gain insights that will help researchers, developers, and data science practitioners accelerate training and iterate faster than ever. Learn (1) best practices for deploying an end-to-end deep learning practice, (2) how the newest DGX systems including DGX Station address the bottlenecks impacting your data science, and (3) how DGX software including optimized deep learning frameworks give your environment a performance advantage over GPU hardware alone. 25 Minute Talk Charlie Boyle - Senior Director, Product Management and Product Marketing, NVIDIA
Add to My Interests
DC7108 - Disruptive Changes in Ophthalmology by Deep Learning Hear about how GPU technology is disrupting the way your eye doctor works and how ophthalmic research is performed today. The rise of electronic medical records in medicine has created mountains of big data, particularly in ophthalmology, where many discrete quantitive clinical elements like visual acuity can be tied to rich imaging datasets. We'll explore the transformative nature that GPU acceleration has played in accelerating clinical research and show real-life examples of deep learning applications to ophthalmology in creating new steps forward in automated diagnoses, image segmentation, and computer-aided diagnoses. 25 Minute Talk Aaron Lee - Assistant Professor, University of Washington
Add to My Interests
DC7231 - Distributed GPU-Enabled TensorFlow Applications at Scale We'll dive deep into large-scale GPU-enabled distributed computing with TensorFlow on an on-premises cluster or the cloud. Since the introduction of distributed TensorFlow, many solutions have been used in an attempt to streamline the cumbersome process of application development and deployment. Having encountered these issues in our own GPU cluster, we present our system architecture and scheduling framework that addresses GPU allocation, containerization, deployment, and scheduling of GPU-enabled TensorFlow applications on a cluster. We will be demonstrating our current use cases that include large-scale distributed training of TensorFlow models and model hyperparameter optimization. 50 Minute Talk Keegan Hines - Machine Learning Engineer, Capital One Financial
Thuc Tran - Machine Learning Engineer, Capital One Financial
Athanassios Kintsakis - Machine Learning Engineer, Capital One Financial
Add to My Interests
DC7241 - Domain Adaptation for Clinical Applications of Ultrasound We'll outline the challenges of training deep neural networks to segment, classify, and detect structures in medical images, and propose practical solutions for ultrasound image analysis. Unlike training neural networks on natural images, the difficulty in the medical domain is in accurately labeling the structures of interest in sufficient quantities, data variation caused by medical conditions, privacy and compliance issues, and correctly defining the requirements based on domain knowledge. We'll present how to train convolutional deep learning network architectures for accurate segmentation of structures in ultrasound images. We'll also show how to use the segmented structures, such as heart chambers, as reliable biomarkers to evaluate health outcomes. 25 Minute Talk Michal Sofka - Scientist, Butterfly Network
Add to My Interests
DC7154 - Effectiveness of Deep Learning Compared to Machine Learning in Applied HealthCare We'll present a use case of applying machine learning and deep learning to the task of imputing/predicting a medical patient diagnosis based on data elements of their member, medical, and pharmacy claims. We'll introduce deep learning approaches, a side by side comparison of machine learning models vs. deep learning models, and illustrate the operation and business value of deep learning models. 25 Minute Talk Julie Zhu - Principal Data Scientist, United Healthcare Group - Optum Technology
Ravishankar Rajagopalan - Engineering Manager – Data Sciences, Optum Technology
Dima Rekesh - Senior Distinguished Engineer, United Healthcare Group - Optum Technology
Add to My Interests
DC7162 - Expanding Man's Mission to Space with VR There are four main areas where NASA uses virtual reality. The Extra-Vehicular Activities training prepares astronauts for space walks. Simplified Aid for EVA Rescue, or SAFER, simulates a situation where astronauts become detached from the shuttle, and have to use a backpack to navigate their way back. NASA also uses VR for robotics operations, relating to the shuttle and space station arm. Finally, zero-g mass handling training helps astronauts get a feel for doing things like manipulating payloads in zero gravity. Everything from the many iterations of headsets to NASA's own graphics engine (dynamic onboard ubiquitous graphics, or DOUG) is produced in-house. The current HMDs they use have 1200x800 resolution, a 120-degree field of view, and no lag. VRs are an important part of how NASA not only trains astronauts before they go into space, but, increasingly, while they're there. 50 Minute Talk Stephen Hunt - Manager, Computer Resources & C&DH Architecture, NASA
Add to My Interests
DC7211 - Exploring Deep Neural Network Architectures for Automated Electron Micrograph Segmentation Learn about our ongoing work at the Laboratory of Cellular Imaging and Macromolecular Biophysics to develop practical, automated 3D segmentation tools for the biological electron microscopy (EM) community. After outlining image segmentation and its role in biological EM, we'll describe present challenges to automating the segmentation of complex biological structures in electron micrographs. To solve those challenges, our lab is developing genenet, a Python package that simplifies the creation, training, and deployment of deep neural networks for segmentation in TensorFlow on multi-GPU and distributed systems. Our session will focus on using manual and algorithmic segmentation network design, and integrating trained networks into EM image analysis workflows. 25 Minute Talk Matthew Guay - Postdoctoral Research Fellow, National Institutes of Health
Add to My Interests
DC7190 - Exploring GPU Inference in the Datacenter New algorithms leverage the algebraic strengths of GPUs far beyond rendering visuals. They unlock opportunities for data analysis leveraging computer vision and artificial neural networks. Earlier this year we set out to investigate the deployment of power-efficient GPUs in commodity hardware. We did not focus on supercomputers, but instead exercised GPUs within a homogeneous set of compute nodes – like those used to scale Apache Hadoop or Apache Spark clusters. Our work focused on inference – deploying models and GPU acceleration for analysis tasks such as feature extraction, identification, and classification – not on training or building models, tasks likely better suited to HPC-class machines. Our experiments investigated applications that aren't feasible at scale on existing CPUs, such as malware detection and object detection in images. We'll cover inference on Tesla P4 GPUs in scale-out architectures, leveraging nvidia-docker, Caffe, Torch, and TensorRT. 50 Minute Talk Drew Farris - Chief Technologist, Booz Allen Hamilton
Add to My Interests
DC7248 - Extreme Computing, Clinical Medicine and GPUs Images and sensors provide crucial information needed to make treatment decisions and machine learning methods are increasingly employed to supplement subjective human image interpretations and to integrate heterogeneous collections of information. We'll describe the rapidly changing landscape of medical images and sensors from both a computing, data, and medical point of view. We'll then do a deep dive in the area of pathology image analytics along with contributions made by deep learning methods to precision medicine and clinical diagnostics. Finally, we'll address the pivotal role of GPUs in supporting all of these computations and describe the roles of GPU-related tools, languages, and libraries in the medical image and sensor analytics. 25 Minute Talk Joel Saltz - Chair and Professor Department of Biomedical Informatics, Stony Brook Medicine and College of Engineering and Applied Sciences Stony Brook
Add to My Interests
DC7225 - Eyeballs to Architecture: How AR and VR are Upending Computer Graphics We'll discuss the ways in which perception, behavior, and uses are forcing us to significantly redefine the graphics pipeline. In the realm of virtual and augmented reality, computer graphics is first and foremost a perceptual stimulus. While we have been quite successful at rendering realistic imagery in real time within the confines of traditional rectangular desk and wall-mounted displays, migrating this technology into immersive visual, auditory, and haptic environments of VR and AR has shown us shortcomings in displays, algorithms, and architectures. The expanded range of applications enabled by AR and VR place additional demands on the quality, comfort, and physical form of graphics devices. 50 Minute Talk Turner Whitted - Researcher, NVIDIA
Add to My Interests
DC7176 - Fast Item Response Theory (IRT) Model Estimation by Using GPUs Learn how to use GPUs to speed up the statistical inference when applying Item Response Theory (IRT) models, which describe the statistical relationships among students' test performances, their latent abilities, and test questions' difficulty levels. IRT has been acting as the cornerstone for many education applications, for example, adaptive computer-based assessments and personalized learnings. How to quickly estimate a large number of parameters of IRT models is a challenging task, especially when facing large-sized educational datasets. We'll introduce how to use a modern probabilistic modeling toolkit, Edward, which uses TensorFlow as its backend for efficiently estimating IRT parameters. Compared to CPUs, we found that GPUs can make IRT parameter-estimation four times faster. 50 Minute Talk Lei Chen - Senior Research Scientist, Educational Testing Service (ETS)
Lei Chen - Principal Scientist, Liulishuo
Add to My Interests
DCL7116 - Full Motion Video Data Curation with TensorFlow Many existing deep learning applications handle video by breaking it apart into individual frames and processing them as static images, completely ignoring the temporal component associated with video. To properly utilize the additional information contained in Full Motion Video (FMV), changes must be made at every step of the deep learning pipeline, starting with data curation. In this lab, we will start with a raw dataset and perform all of the necessary data curation steps required to successfully apply Deep Neural Networks (DNN) to FMV. We will go through the inherent complexities associated with FMV when preparing data for training DNNs, checking the quality of training, testing the quality of results and visualizing the final output. In this process, you will learn to encode and decode video, perform large scale video transformation operations and performance optimizations required for high speed inference on live video. Pre-requisites: Nvidia's Deep Learning Fundamentals Workshop OR a working understanding of Deep Learning Fundamentals such as: Some familiarity deep learning but no background with video is required Comfortable with scripting, dealing with data files and TensorFlow 120 Minutes Instructor-Led Lab Michael Demoret - Architect, Solutions, NVIDIA
Add to My Interests
DC7147 - GPU-Accelerated Data Science with GOAI Modern data science workflows typically combine multiple tools, APIs, and technologies. Therefore, shared standard data structures are essential to building an efficient workflow. Without a shared data structure, workflows must copy and convert data when going from one tool to another. Data movement and conversion introduces latency and unneeded complexity.  The GPU Open Analytics Initiative (GOAI) is an effort by Continuum Analytics, H2O.ai, MapD, BlazingDB, Graphistry and the Gunrock project (UC Davis) with the aim to create open frameworks that allow developers and data scientists to build applications using standard data formats and APIs on GPUs.  This will allow data scientist to seamlessly move between tools at GPU-accelerated speeds. This talk will present the various components of the GOAI framework and walk through a typical data science task using GOAI. 25 Minute Talk Brad Rees - Senior Solution Architect, NVIDIA
Add to My Interests
DC7181 - GPUs And Sovereign Clouds: Empowering the Public Sector Frame is a revolutionary service that lets you move Windows and Linux applications to the cloud and access them from any device. All you need is a browser ­– no plugins or additional hardware required. With Frame, you can quickly publish applications to any Azure or AWS datacenter globally - including sovereign regions such as Azure Government. Frame is optimized for NVIDIA graphics, so even the most complex visual applications run smoothly in a browser at up to 60FPS and with multiple monitors. We'll share how Frame is empowering the Public Sector with powerful cloud hosted app delivery solutions using NVIDIA GPUs - simplifying management and speeding up access to everything from GIS software to productivity tools. We'll show you everything you need to know to run your applications today, without wasting time or dealing with the hassles of legacy remoting technologies. 25 Minute Talk Jason Holloway - Head of Public Sector, Frame
Add to My Interests
DC7149 - Graph Centric AI for Cybersecurity Large enterprise networks face the daily challenge of cyberattacks, which originate from software and hardware vulnerabilities and result in data theft, service interruption, and monetary loss. To address this challenge, we've developed a set of graph-based machine learning techniques for accelerating threat detection on GPUs. We'll present our research on graph-centric AI that can be used to discover malicious actions in time to prevent irreversible damage to the systems. In the era of big data, these techniques help us to have a deep understanding of critical relationships in computer systems, social networks, and IoT, which is essential in different industry segments, including defense, software, finance, e-commerce, and healthcare. 50 Minute Talk Howie Huang - Associate Professor, The George Washington University
Add to My Interests
DCL7112 - Hands-on with Driverless AI In this lab you'll learn how to install and start Driverless AI, the automated Kaggle Grandmaster in-a-box software, on a multi GPU box. We'll go through the full end to end workflow of taking a dataset, cleaning it and performing EDA using Auto Viz and then running it through a bunch of automated recipes for feature engineering and model building to create a really accurate model. We'll then analyze and interpret the results of the model using the machine learning interpretability toolkit. And finally we'll deploy the best model and pipeline with Driverless AI's scoring service. Driverless AI uses the power of GPUs to achieve almost 40x speedups on algorithms that it turn allow it run thousands of iterations and find the best model. 120 Minutes Instructor-Led Lab Arno Candel - CTO, H2O.ai
Add to My Interests
DC7240 - Harnessing AI in Healthcare As computers outperform humans at complex cognitive tasks, disruptive innovation will increasingly remap the familiar with waves of creative destruction. And in healthcare, nowhere is this more apparent or imminent than at the crossroads of Radiology and the emerging field of Clinical Data Science. As leaders in our field, we must shepherd the innovations of cognitive computing by defining its role within diagnostic imaging, while first and foremost ensuring the continued safety of our patients. If we are dismissive, defensive or self-motivated - industry, payers and provider entities will innovate around us achieving different forms of disruption, optimized to serve their own needs. To maintain our leadership position, as we enter the era of machine learning, it is essential that we serve our patients by directly managing the use of clinical data science towards the improvement of care—a position which will only strengthen our relevance in the care process. 60 Minute Keynote Keith Dreyer - Vice Chairman of Radiology, Massachusetts General Hospital, , Massachusetts General Hospital/Harvard Medical School
Add to My Interests
DC7253 - Harnessing Machine Learning to Enable Global Discovery at Scale There has been an explosion in the number of commercially available satellite images produced every day. The democratization of space technology has catalyzed a revolution of the commercial space industry, which is now rapidly transforming from one company imaging five million km2 a day to around 10 companies imaging 200 million km2 a day; and from constellations of a handful of satellites to constellations of hundreds of satellites. Today, the number of images being generated by this rapidly evolving commercial space industry far exceeds human scales. We'll explore how we can harness machine learning to enable global discovery at scale. Fueled by advancements in machine learning algorithms, GPUs, and the availability of labeled datasets, we have dramatically improved our ability to extract insight from this explosion of data to understand changes across the globe. 50 Minute Talk Mikel Rodriguez - Research Scientist , MITRE
Add to My Interests
DC7119 - High Temporal Resolution ISR Data and Synergistic Decisions through Multi-Stream Deep Learning We'll cover research Harris has performed in two applications of deep learning leveraging NVIDIA GPUs. The first topic pertains to high temporal resolution datasets. Many of the "smallsat" providers are offering imagery collections at extremely high temporal resolution. Deep learning provides the ability to automate much of the routine analysis of this data. The second topic will walk through several use cases Harris has developed harnessing multiple input sources for deep learning. Deep learning has been applied to automatic feature extraction for a variety of individual data types such as imagery, point clouds, and video. One of the less explored benefits of deep learning in remote sensing data is the ability to incorporate multiple streams of data into the same neural network, revealing more information than either of the individual data types can provide alone. 50 Minute Talk William Rorrer - Program Manager, Harris
Add to My Interests
DC7226 - How AI is making Your City Smarter and Safer Our cities are increasingly challenged to bring intelligence and safety to their citizens' lives. New technologies using AI are promising to change the way cities operate. How will citizens benefit? What are the policies and investments necessary to bring this about as soon as possible? 50 Minute Panel Naveen Lambda - Director of Analytics, Grant Thornton
Richard Kidd - Deputy Assistant Secretary, U.S. Army
Archana Vemulapalli - CTO, Office of the Chief Technology Officer Office of Communications
Milind Naphade - Chief Technology Officer, AI Cities, NVIDIA
Russell Brooks - Director of Smart Cities for Transportation, DOT
Darrell Issa - U.S. Representative for California, House of Representatives
John Garofolo - Senior Advisor, Information Access Programs, National Institute of Standards and Technology
Milind Naphade - Chief Technology Officer, AI Cities, NVIDIA
Add to My Interests
DC7116 - How an Architectural Design Firm Leverages Virtual GPU Technology for Global Collaboration Learn the benefits that virtualization provides for an architecture and engineering design firm, along with the journey through the advancements in virtualization technology it took to finally meet the graphics-intensive needs of our design software. We'll share our experiences in how virtualization allows a large company, with over 15 offices and 1,000 people worldwide, to collaborate and work as a single firm. We'll show some cost comparisons with virtualization, along with their management benefits and requirements. We'll also look at the methods we used to set and test metrics specific to our requirements, and follow the results of those metrics through the changes in graphics virtualization technology. 25 Minute Talk Andrew Schilling - Chief Infrastructure Officer, Cannon Design
Jimmy Rotella - Digital Practice Director, CannonDesign
Add to My Interests
DC7250 - How Deep Learning Reverses Resolution-Degrading Effects of Conventional Video Capture We'll showcase both the technology and use-cases for applying convolutional neural networks and GPUs to reverse the resolution-degrading effects of optical blur and sensor sampling, in order to reconstruct color video to nine times its captured pixel density. 25 Minute Talk Michael Korkin - CEO, Entropix
Dwight Linden - COO, Entropix
Nathan Wheeler - Chief Product Officer, Entropix
Add to My Interests
DC7244 - How Startups and Other Emerging Technology Providers Can Work with Federal Agencies Although the major federal agencies have substantial technology budgets, they pale in comparison to the aggregate amount of funding going into startups around the world. Agencies are increasingly looking to outside vendors, especially startups, to provide advancements in critical technologies relating to AI, cybersecurity, and other mission-critical areas. Hear from a panel of experts, including the director of science and technology at the CIA, about how and why agencies are increasingly working with startups and emerging technologies. How can startups best position themselves to do business with the federal government? What areas of technology are currently the most attractive? What are the key issues that need to be addressed? 50 Minute Panel James Crawford - Founder & CEO, Orbital Insight
George Hoyem - Manging Partner, In-Q-Tel
George Hoyem - Manging Partner, In-Q-Tel
Dawn Meyerriecks - Deputy Director for Science and Technology, CIA
Todd Mostak - Founder & CEO, MapD
Jeff Herbst - VP Business Development, NVIDIA
Jeff Herbst - VP Business Development, NVIDIA
Add to My Interests
DCL7105 - Image Classification with DIGITS Learn how to leverage deep neural networks (DNN) within the deep learning workflow to solve a real-world image classification problem using NVIDIA DIGITS. You will walk through the process of data preparation, model definition, model training and troubleshooting. You will use validation data to test and try different strategies for improving model performance using GPUs. On completion of this lab, you will be able to use DIGITS to train a DNN on your own image classification application. 120 Minutes Instructor-Led Lab Lawrence Brown - Certified Instructor, NVIDIA
Add to My Interests
DCL7101 - Image Classification with TensorFlow: Radiomics - 1p19q Chromosome Status Classification Thanks to work being performed at Mayo Clinic, approaches using deep learning techniques to detect Radiomics from MRI imaging can lead to more effective treatments and yield better health outcomes for patients with brain tumors. Radiogenomics, specifically Imaging Genomics, refers to the correlation between cancer imaging features and gene expression. Imaging Genomics (Radiomics) can be used to create biomarkers that identify the genomics of a disease without the use of an invasive biopsy. The focus of this lab is detection of the 1p19q co-deletion biomarker using deep learning - specifically convolutional neural networks – using Keras and TensorFlow. What is remarkable about this research and lab is the novelty and promising results of utilizing deep learning to predict Radiomics. 120 Minutes Instructor-Led Lab Hoo Chang Shin - Certified Instructor, NVIDIA
Add to My Interests
DC7223 - Improving Security through AI - Next Generation of Cybersecurity Deep learning and AI will revolutionize cybersecurity by dramatically improving detection and intrusion capabilities. However, we can't completely eliminate cyberthreats. We'll discuss how new AI technologies are working to enhance security, while also examining potential risks. We'll also discuss what role regulation should play in ensuring private institutions appropriately strike the right risk balance, and how government and industry are working together to combat cybercrimes. 50 Minute Panel Leo Meyerovich - CEO, Graphistry, Inc.
Iain Cunningham - VP of Intellectual Property and Cybersecurity, NVIDIA
Ashok Pinto - Chief Investigative Counsel & Policy Director Committee on Commerce, Committee on Commerce, Science, & Transportation, U.S. Senate
Christian Fjeld - Senior Counsel for Consumer Protection, Product Safety, Senate Committ, Senate
John Ratcliffe - Chairman of the Cybersecurity and Infrastructure Protection Subcommitt, US Government
Peter Guerra - VP Chief Data Scientist, Booz Allen
Iain Cunningham - VP of Intellectual Property and Cybersecurity, NVIDIA
Add to My Interests
DC7201 - Inference in the Data Center on Tesla In deep learning, inference is where neural networks deliver insights. What started with images has quickly expanded to include speech, neuro-linguistic programming, recommender systems, and video. As datasets get bigger, networks get deeper and more complex, and latency requirements get tighter, GPUs are the ideal platform to accelerate these workloads, both for high-batch and low-latency use-cases. Learn how inference gets done on GPUs, get the latest on updates on the software stack for inference — including the TensorRT inference engine — and DeepStream SDK, and recommendations on choosing the right GPU for running inference workloads. 25 Minute Talk Chris Gottbrath - Senior Manager, NVIDIA
Add to My Interests
DC7169 - Insights From Room-Scale Virtual Reality Version of a Classic Motor Learning Paradigms The serial response time task (SRTT) is a classic motor sequence learning paradigm that has been used to study human motor skill learning for four decades. Typically studies use a key-press response and a fixed sequence of movements, but variants will also involve reaching and more complex sequence structures. In all versions of this task, a level of abstraction is required as none of the movements in these lab-based tasks correspond to movements in a daily real-life environment. However, along with improvements in head-mounted displays and hand-held trackers, powerful GPUs and realistic simulation of physics in virtual environments have made it possible to create an immersive, room-scale version of this classic task that allows natural movements that have direct correspondence to movements in a daily-real life environment, and track response time and trajectory information. 50 Minute Talk Sunbin Song - Research Scientist, NIH NINDS
Add to My Interests
DC7131 - Intelligent Process Automation with Jetson-TX2 We'll focus on engineering and building an intelligent process automation solution with Angular, Activiti, TensorFlow, and Jetson TX2. While workflow and BPM solutions have long been a commodity capability, the reality is that many business processes and their decision making have to be supplemented with sufficient context and strong analytics. Without the context and analytics, some decisions get delayed or never get made thus defeating the purpose of automation. Even the most basic digitization effort attempting to do away with paper and dealing with OCR requires a significant amount of context and analytics to be done correctly. We'll show how to build a lightweight "intelligent process automation" capability by assembling open source components such as Angular, Activiti, and TensorFlow, where the OCR processing runs on the Jetson TX2 system. We'll describe the modular architecture and the actual code in assembling a fully functional product, and we'll share learned lessons. 25 Minute Talk Murali Kaundinya - Director, Merck
Add to My Interests
DC7212 - Interpretable AI: Not Just For Regulators We'll share several approaches beyond the error measures and assessment plots typically used to interpret deep learning and machine learning models and results. We'll demonstrate a full featured interpretability toolkit using H2O's Driverless AI. While understanding and trusting models and their results is a hallmark of good (data) science, model interpretability is a serious legal mandate in the regulated verticals of banking, insurance, and other industries. Moreover, scientists, physicians, researchers, and humans in general have the right to understand and trust the models and modeling results that affect their work and their lives. Today, many are embracing deep learning and machine learning techniques, but what happens when people want to explain these impactful, complex technologies or when these technologies inevitably make mistakes? 25 Minute Talk Patrick Hall - Director of Data Science, H2O.ai
Add to My Interests
DC7117 - Introducing NVIDIA Holodeck Bring your ideas to life with NVIDIA Holodeck, the world's first intelligent, photorealistic, and collaborative virtual reality platform. With Holodeck designers will be able to visualize large, highly detailed models and explore them in photo-real fidelity — in real-time. Design teams can collaborate on these complex models remotely to discover new ideas, streamline reviews, and minimize costly physical prototyping. Holodeck even promises to tap into AI to accelerate design workflows and complex simulations. Come hear the talk and then experience Holodeck demos in the VR Village! 50 Minute Talk David Weinstein - Director Pro VR, NVIDIA
Add to My Interests
DC7182 - Keep Those GPUs Busy: How to Plan Storage for Deep Learning We'll dive into the critical role of storage in deep learning infrastructure. When properly sized for capacity and performance, storage eliminates time-consuming data copies and provides the throughput and IOPS to keep both GPUs busy and data scientists productive. Learn about the storage requirements and performance characteristics of model training, interactive model experimentation, data cleaning and transformation, and data ingest. You'll hear practical lessons, such as helping Pure Storage customers plan deep learning infrastructure in areas that include self-driving cars, social media, and image recognition. 50 Minute Talk Igor Ostrovsky - Software Architect, Pure Storage
Add to My Interests
DC7109 - Learn how to Provide High Performance Public Safety VDI's with NVIDIA GPUs We'll talk about the current performance limitations with VDI and its associated slow adoption; solutions to these limitations turning VDIs into the preferred medium for your environment; creating a better than PC experience with excellent manageability; discovering the pitfalls of various system configurations and designs; and exploring the additional advantages of a VDI deployment for any environment. 25 Minute Talk Cory Smith - CIO/CTO, City of Davenport, Iowa
Add to My Interests
DC7237 - Learning Semantic Video Captioning Using Data Generated with Grand Theft Auto Learn an effective strategy for training deep video captioning and semantic search of overhead scenes, taking advantage of the photo-realism and ubiquitous access to high-level semantic information in a game like Grand Theft Auto. We train video captioning models and are surprised by how well their performance transfers to real drone, security camera, and even infrared footage in the face of poor video quality, compression artifacts, occlusion, and clutter. The speed of NVIDIA GPUs makes it possible to run the game engine with computationally heavy realism-augmenting mods, train video captioning models quickly, run captioning models faster than real-time video, and leverage commercially trained image models to quickly perform semantic search-by-example over large video datasets at scale. 25 Minute Talk Kolia Sadeghi - Applied Mathematician, CCRi
Alexander Polis - Data Scientist, CCRi
Alexander Polis - Data Scientist, CCRi
Add to My Interests
DC7142 - Leveraging Deep Learning and Video Analysis in Law Enforcement Body-worn cameras have proven to strengthen trust and accountability between law enforcement agencies and the communities they serve. However, large-scale use of body-worn cameras has generated massive amounts of data, which is practically impossible for these agencies to use effectively. This has led to significant, and unproductive, time spent manually analyzing data. Axon Research is using the latest advances in deep learning and GPU acceleration to enable increased efficiency across the body-worn camera continuum by accelerating the many manual, time-consuming workflows in public safety, such as redacting footage in response to a public request. Attendees will hear the potential impact of large-scale deep learning on law enforcement and public safety information management. 50 Minute Talk Sanchit Arora - Lead Researcher, Axon Research Group
Add to My Interests
DC7134 - Machine Learning for Malware Analysis We'll explore the ways in which machine learning can be applied to the malware detection and prevention problem. We discuss portable executables, the features derived therefrom, and the ways in which GPU-trained deep neural networks and graph kernels can be used to discriminate between malicious and benign samples. 50 Minute Talk Michael Slawinski - Data Scientist, Cylance
Add to My Interests
DC7233 - Machine Perception at the Tactical Edge The rapid deployment of video sensors across multiple platforms, such as security cameras, unmanned aerial vehicles, and satellites, has resulted in information overload, outpacing analyst ability to effectively use the capability. State-of-the-art processing, exploitation, and dissemination systems primarily focus on forensic use of stored video for identification and tracking of objects and subjects of interest. Not many capabilities exist to deploy and monitor in real-time 100s and 1,000s sensors in cities, bases, airports, and similar venues. Using cloud deep learning services or APIs (for example, Amazon Rekognition) is not only cost prohibitive, but also presents challenges to organizations with sensitive or classified data. We'll discuss various efforts at the Johns Hopkins University Applied Physics Laboratory to develop inexpensive, low size, weight, and power (SWaP) real-time automatic target recognition systems. 50 Minute Talk Pedro Rodriguez - Senior Research Scientist, JHU Applied Physics Laboratory
Add to My Interests
DC7113 - Malware Detection by Eating a Whole EXE Malware is a problem of increasing impact as our society becomes more dependent on an infrastructure of networked computers. Detecting malware is not a trivial task, and numerous companies and startups are researching ways that machine learning could help solve this problem. At the Laboratory for Physical Sciences, we've been researching new ways of applying deep learning to this problem. In contrast to most works in the subject area, ours has been done while minimizing the amount of domain knowledge used. We'll share our results in training a neural network that processes an entire binary from scratch, giving us a malware detection system that requires no knowledge to train. This gives our approach considerable flexibility and can be repurposed to new domains with minimal effort. However, it also poses a number of machine learning and engineering challenges to overcome. 50 Minute Talk Jon Barker - Senior Research Scientist, NVIDIA
Robert Brandon - Security Researcher, Booz Allen Hamilton
Edward Raff - Senior Lead Scientist , Booz Allen Hamilton
Jared Sylvester - Lead Scientist , Booz Allen Hamilton
Add to My Interests
DCL7104 - Medical Image Analysis with R and MXNet In this lab you will use the deep learning framework MXNet to train a CNN to infer the volume of the left ventricle of the human heart from a time-series of volumetric MRI data. You will learn how to extend the canonical 2D CNN to be applied to this more complex data and how to directly predict the ventricle volume rather than generating an image classification. In addition to the standard Python API, you will also see how to use MXNet through R, which is an important data science platform in the medical research community. 120 Minutes Instructor-Led Lab Robert Keating - Certified Instructor, NVIDIA
Add to My Interests
DCL7103 - Modeling Time Series Data with Recurrent Neural Networks in Keras One important area of current research is the use of deep neural networks to classify or forecast time-series data. Time-series data is produced in large volumes from sensors in a variety of application domains including Internet of Things (IoT), cyber security, data center management and medical patient care. In this lab, you will learn how to create training and testing datasets using electronic health records in HDF5 (hierarchical data format version five) and prepare datasets for use with recurrent neural networks (RNNs), which allows modeling of very complex data sequences. You will then construct a long-short term memory model (LSTM), a specific RNN architecture, using the Keras library running on top of Theano to evaluate model performance against baseline data. 120 Minutes Instructor-Led Lab Robert Keating - Certified Instructor, NVIDIA
Add to My Interests
DC7164 - M&S 4D Technology Cross-Pollination Trends: CUDA for Live Sensor Processing M&S 4D technologies are rapidly cross-pollinating beyond training and simulation. After initially pushing into mission rehearsal, they are now transitioning as an innovative force into live applications as well for both manned and unmanned ISR and Strike environments. NVIDIA has enabled advances in multi-spectral rendering, geospatial databases, and real-time computed fluid dynamics. The crossroads of 4D spectral rendering technology, 3D high-ground truth global databases and real-time, multi-sensor fusion might lead the way in future theaters of war. One example in our CUDA toolbox saw 150,000x speedup from CPU prototype to optimized GPU implementation. We'll present techniques of instantaneous image decimation; CDF via warp shuffle; designing XY-separable kernels and their intermediate data; sliding window tradeoffs; and solution approximations for minimal context switching. 50 Minute Talk Javier Castellar - Co-founder & VP of Business Development, Aechelon Technology Inc.
Sarah Kabala - High-Performance Graphics Engineer, Aechelon Technology, Inc.
Add to My Interests
DCL7117 - Multi-object Tracking in Full Motion Videos using Spatial and Temporal Features In video data, there are often multiple objects moving in a scene at different speeds and various durations in time. By exploiting the spatial and temporal aspects of the data, more advanced deep learning analytics, like multi-object tracking, become possible. Upon completion of the pre-requisite lab, you will be able to curate and organize training data from video sources for these types of deep learning training tasks. In this lab you will take a deep technical dive into the training of a temporally-aware deep learning model for multi-object target tracking. We will explore approaches in training and optimization for improving the tracking model performance. Pre-requisites Full Motion Video Data Curation with TensorFlow 120 Minutes Instructor-Led Lab May Casterline - DLI Certified Instructor, NVIDIA
Add to My Interests
DC7121 - Multi-User VR Solutions for Enterprise Deployment Deploying PC-based virtual reality solutions throughout the enterprise poses challenges beyond the typical consumer model of one PC driving one headset for one user at one location. For consumer VR inside the home, the primary user typically owns, maintains, and controls access to both the PC and physical location. Business needs are different. As the number of locations or simultaneous users at each location increases, manageability becomes difficult and unwieldy. Enterprise requirements come into play such as deployment to temporary locations with limited setup/pack up time, limited physical space, robustness, scaling to many concurrent users, multi-user collaboration, remote IT management, configuration control, and system image replication. We'll introduce an experimental approach to multi-user VR deployment based on virtualization techniques that aims to address these enterprise use-case requirements. 50 Minute Talk Tom Kaye - Sr. Solution Architect, NVIDIA
Add to My Interests
DCL7107 - Neural Network Deployment with DIGITS and TensorRT This lab will show three approaches for deployment. The first approach is to directly use inference functionality within a deep learning framework, in this case NVIDIA DIGITS and Caffe. The second approach is to integrate inference within a custom application by using a deep learning framework API, again using Caffe, but this time through its Python API. The final approach is to use the NVIDIA TensorRT™, which will automatically create an optimized inference run-time from a trained Caffe model and network description file. In this lab, you will learn about the role of batch size in inference performance, as well as various optimizations that can be made in the inference process. You will also explore inference for a variety of different DNN architectures trained in other DLI labs. 120 Minutes Instructor-Led Lab Lawrence Brown - Certified Instructor, NVIDIA
Add to My Interests
DC7238 - New Video Search Capabilities for Public Safety through Intelligent Video Analytics and Deep Learning For security teams working to ensure public safety, the ability to minimize incident response time and speed forensic investigations is critical. We'll discuss a new end-to-end, deep learning, and GPU-reliant architecture and video search engine for video data being deployed to solve this. 25 Minute Talk Mahesh Sapharishi - Chief Technology Officer, Avigilon
Add to My Interests
DC7234 - Next Steps to Creating Personalized and Adaptive Virtual Reality/Augmented Reality Applications Recent advances in virtual reality/augmented reality have made the future into the present. Disruptive technologies that could only have been imagined at the turn of the 21st century are now poised to become accessible to a diverse group and in a wide range of settings. But these advances raise a whole new set of challenges. The National Science Foundation recently hosted a VR/AR workshop to bring together industry, academia, and government stakeholders to envision the future of the field and explore challenges. The workshop focused on the challenges of achieving individual personalization and adaptation in VR/AR applications. The workshop identified a range of needs, including: methods for personalizing the human-machine interface; accurate and unobtrusive approaches for identifying relevant internal and physical user states; techniques to support dynamic content generation for user adaptation; and new methods for addressing and overcoming cybersickness. The workshop also stressed the necessity of developing user-centric benchmarks to guide personalization; guidelines for how and when to exploit and optimize spatial cognition; and the knowledge of the roles of social, behavioral, and cognitive norms and patterns in the development of next generation VR/AR. We'll share these key findings and the necessary next steps to create personalized VR/AR across a range of different applications. 50 Minute Talk Wendy J. Nilsen - Program Director, National Science Foundation
Add to My Interests
DC7125 - NVIDIA Isaac – Virtual Environments for Robot AI We'll describe NVIDIA's Isaac system - a virtual training environment for artificially intelligent robots. Isaac exists at the intersection between computer graphics and AI, leveraging improvements in real-time graphics visualization to provide a realistic simulation environment in which robotic vision and interaction systems can be trained. 50 Minute Talk Gavriel State - Research Scientist, NVIDIA
Add to My Interests
DC7120 - NVIDIA Tools and SDKs for Training and Simulation To complement the world's most powerful GPUs, NVIDIA develops a wide range of software tools and SDKs that can be used to make better training and simulation systems. These include support for multi-GPU configurations, VR, advanced rendering, high performance computing, virtualization, and deep learning. 50 Minute Talk Tim Woodard - Sr. Solutions Architect, NVIDIA
Add to My Interests
DCL7106 - Object Detection with DIGITS This lab explores three approaches to identify a specific feature within an image. Each approach is measured in relation to three metrics: model training time, model accuracy and speed of detection during deployment. On completion of this lab, you will understand the merits of each approach and learn how to detect objects using neural networks trained on NVIDIA DIGITS using real-world datasets. 120 Minutes Instructor-Led Lab Lawrence Brown - Certified Instructor, NVIDIA
Add to My Interests
DC7242 - Nutanix AHV & NVIDIA: Putting the V in vGPU! Nvidia and Nutanix have collaborated at an engineering level to bring together the power or of 1-click enterprise cloud OS and nVidia's GPU technologies. While Nutanix platforms have supported nVidia GPU cards for a while now, with the upcoming release of Nutanix Acropolis software, we take that partnership to the next level with graphics virtualization support for Nvidia vGPUs on Nutanix's embedded hypervisor – Acropolis (AHV). Customers will be able to immediately leverage this not just for app and desktop virtualization applications (eg. Citrix), but for also high performance computing applications. Attend this session to learn more about Nutanix's hyperconverged infrastructure solution, how it can simplify datacenter infrastructure management challenges, Nutanix's embedded Acropolis hypervisor, and its support for Nvidia vGPU integration. 25 Minute Talk Kelly Oliver - Solutions Architect, Nutanix
Saveen Pakala - TBA, Nutanix
Add to My Interests
DC7102 - Optimizing, Profiling, and Deploying TensorFlow AI Models in Production with GPUs Using the latest advancements from TensorFlow, including the accelerated linear algebra (XLA) framework, JIT/AOT compiler, and graph transform tool, Chris will demonstrate how to optimize, profile, and deploy TensorFlow models in GPU-based production environments. This talk is 100 percent demo based with open source tools and completely reproducible through Docker on your own GPU cluster. We'll provide a GPU for every attendee to follow along during the talk. 50 Minute Talk Chris Fregly - Founder and Research Engineer, PipelineAI
Add to My Interests
DC7180 - Practical Applications of State-of-the-Art Neural Networks to Surveillance Systems We'll discuss practical applications of state-of-the-art convolutional neural networks (CNNs) to the challenges of surveillance systems. We'll discuss modifications to a current state-of-the-art neural network to improve detection of small targets while maintaining detection performance on larger targets. We'll then discuss a novel sensor-fusion method to improve small target localization and provide track classification to a fused common operating picture. We'll present results from the MS-COCO dataset and from data collected on our own testbed, which recorded small quadcopters in flight using a commercial surveillance camera. Finally, we'll discuss how our novel CNN detection and sensor fusion algorithms can be used to enable surveillance security applications. Sponsor: DHS S&T. 50 Minute Talk Peter Morales - Associate Staff, MIT Lincoln Laboratory
Virginia Goodwin - Associate Staff, MIT Lincoln Laboratory
Add to My Interests
DC7159 - Precision Healthcare As the industry and government collects massive amounts of data to help provide faster and more accurate clinical care, the gap in managing and exchanging these data types remains a challenge for the industry. Key highlights of this panel discussion will include how AI can advance treatment and prevention, what scientific and regulatory hurdles remain for industry success, and how Congress can best address possible privacy and security issues. 50 Minute Panel Nathan Hubbard - Director Digital and Personalized Healthcare Partnering, Roche Partnering
Thomas Fuchs - Associate Professor, Memorial Sloan Kettering Cancer Center
Kyle Farh - Director, Data Sciences, Illumina
Vern De Biasi - Head of Emerging Platforms, GSK
Andrea DeSouza - Global Business Development Lead, NVIDIA
Andrea DeSouza - Global Business Development Lead, NVIDIA
Add to My Interests
DC7194 - Road Network Detection and Routing Via Satellite Imagery Determining optimal routing paths is at the heart of many humanitarian and military challenges, and also crucial for autonomous vehicles. In areas of low population density or undergoing rapid change (for example, natural disasters), commercial or open source mapping products are often inadequate. The rapid revisit rates of satellite imaging constellations have the potential to alleviate this capability gap, if road networks can be inferred directly from imagery. We demonstrate techniques for extracting the physical and logical topology of road networks from satellite imagery via computer vision and graph theory techniques; the road network graph structure inferred from our algorithms can be used directly for routing. We also develop a new metric based upon shortest path algorithms for measuring network similarity, and show how road network inference performance varies between differing scenes and environments. 25 Minute Talk Adam Van Etten - Senior Research Scientist, In-Q-Tel
Add to My Interests
DC7133 - Scaling Event Data Investigations with GPU Visual Graph Analytics This talk shares how GPUs are enabling investigation teams to answer tough event data questions around progression, scope, root cause, patterns, and anomalies. Most of these problems are equivalent to analyzing graphs for correlations. Drawing examples from areas like cybersecurity, we share how to think about event data as a graph problem, and how scaling graph visualization and analytics with GPUs is enabling investigation teams to unlocks new insights and workflows. In particular, we show: -- Correlating events with hypergraphs, such as for killchain analysis and identifying shell companies -- Visualizing large graphs with GPU rendering and smart designs -- Fast analytics through GPU cloud computing -- Combining the above into daily operational workflows for data scientists, developers, and front-line analysts We draw examples from Graphistry's work in the Fortune 500 and the US government. 25 Minute Talk Leo Meyerovich - CEO, Graphistry, Inc.
Add to My Interests
DC7207 - Simulating Vestibular Disturbances in Astronauts With Optical Stimuli Using Virtual Reality We'll show how optical stimuli can be used to induce vection without causing significant motion sickness for the purpose of simulating disturbances in the vestibular organs residing within the inner ear. We'll cover biological processes required for balance, how those processes are disrupted by transitioning between different gravitational environments, and how similar results can be obtained using a VR-based visual method. We'll then explain how the system can be used as a sensorimotor function assessor and countermeasure tool, and the results of a human performance study leveraging it. Learn how this capability could help train astronauts, treat medical conditions, and aerospace applications, and how NVIDIA's Falcor engine helped fast-track development. 50 Minute Talk Matthew Noyes - Aerospace Technologist, National Aeronautics and Space Administration
Add to My Interests
DC7197 - SpaceNet: Accelerating Automated Mapping with Deep Learning and Labeled Satellite Imagery We'll explain the SpaceNet Challenge Round Two results for automated building footprint extraction and preview future challenges. SpaceNet is an online repository of openly available satellite imagery, co-registered map data to train algorithms, and a series of prize challenges designed to accelerate innovation in machine learning. This first of its kind, open innovation project for the geospatial industry launched in 2016 as a collaboration between Amazon Web Services, CosmiQ Works, DigitalGlobe, and NVIDIA. SpaceNet Round Two incorporated enhancements from previous competitions, including the release of high-resolution imagery from DigitalGlobe's WorldView-3 satellite and improved labeled training data. The winning solution used deep learning to achieve superior performance for building mapping. Such algorithms have the potential to assist in updating maps and can help end-users — such as first responders — to direct resources more effectively. 25 Minute Talk Todd M. Bacastow - Director, Strategic Alliances / SpaceNet lead, DigitalGlobe|Radiant
David Lindenbaum - Principal Engineer, CosmiQ Works
Add to My Interests
DC7173 - Speech from a Distance: An Open Source Dataset for the Audio Community Speech is the pervasive media by which we communicate the most quickly and efficiently. Recent deep learning efforts enabled by quick GPU processing have shown progress in recognizing speech, speakers, and discerning noise from signal. Unfortunately, current methods of mixing noise and speech for training purposes in deep learning are unrealistic and stymie the effective training of machine learning algorithms. Lab41/In-Q-Tel, in conjunction with NVIDIA, Doppler Labs, SRI International, and Amazon, is working to bring the data to the researchers. This data will be oriented in a manner that reflects real physical properties in the acoustics of a room. Most importantly, we intend to release this dataset to the open source creative commons with no strings attached. As part of this release, Lab41 will present novel research in the area of signal processing for benchmarking for use in challenges related to the dataset. 25 Minute Talk Cory Stephenson - Data Scientist, In-Q-Tel
Add to My Interests
DC7136 - The DOE and NCI Partnership on Precision Oncology and the Cancer Moonshot The Cancer Moonshot was established in 2016 with the goal to double the rate of progress in cancer research -- to do in five years what normally would take 10. A major area for the acceleration of progress is the strategy to use modeling, simulation, and machine learning to advance our understanding of cancer biology and to integrate what is known into predictive models that can inform research and guide therapeutic developments. In 2015, the U.S. Department of Energy formed a collaboration with the National Cancer Institute for the joint development of advanced computing solutions for cancer. 50 Minute Talk Rick Stevens - Associate Lab Director, Argonne National Labs
Add to My Interests
DC7158 - The Future of Driving - Paving the Way for Self-Driving Cars Using cutting-edge AI technologies, self-driving cars have the potential to significantly reduce traffic fatalities, improve transportation mobility and accessibility, and increase productivity. To realize the promise of these many benefits, public policy must allow innovation to flourish by, for example, removing barriers to testing and clarifying state and federal regulatory authority. During this panel, we'll hear from top policymakers and industry representatives on what key policies need to be enacted to advance the deployment of self-driving cars, and how all stakeholders are working together to further that goal. 50 Minute Panel Cherilyn Pascoe - Professional Staff Member and Investigator , U.S. Senate Committee on Commerce, Science, and Transportation
Brad Stertz - Director, Audi Government Affairs, Audi
Nat Beuse - Associate Administrator for Vehicle Safety Research , National Highway Traffic Safety Administration (NHTSA)
Michelle Ash - Chief Consumer Protection Counsel at U.S. House of Representatives, U.S. House of Representatives
Danny Shapiro - Senior Director Automotive, NVIDIA
Bert Kaufman - Head, Corporate and Regulatory Affair, Zoox
Danny Shapiro - Senior Director Automotive, NVIDIA
Add to My Interests
DC7130 - The Impact of Deep Learning on Radiology - 2017 Update Major advances in computer science are beginning to have an impact on radiology. The rapid achievements in performance for object detection in natural images have enabled these impacts. There has been an explosion of research interest and number of publications regarding the use of deep learning in radiology. We'll show examples of how deep learning has led to major performance improvements in radiology image analysis, including image segmentation and computer-aided diagnosis. 25 Minute Talk Ronald Summers - Senior Investigator, NIH Clinical Center
Add to My Interests
DC7189 - The Impact of GPUs in Geovisualization for Government Due to the explosion of geospatial data from sensors, smart phones, social media, drones and other vehicles of transportation, new opportunities are arising for government agencies to leverage GPUs to analyze and visualize geospatial data. We'll discuss how GPU-accelerated analytics can be harnessed to extract the fastest insights possible from geospatial data, and how this technology is enabling federal agencies to take faster action and make more informed decisions around issues of security, voting, campaigns, political donations, public services and more. 25 Minute Talk Todd Mostak - CEO & Founder, MapD
Add to My Interests
DC7183 - The Latest in Brain Imagery Using GPU's Learn how NLP Logix and the Mayo Clinic developed a training procedure for a system that will automatically identify and segment white matter hyperintensities (WMH) in T1 and FLAIR MRI sequences using GPUs. This work was done for the White Matter Hyperintensity Segmentation Challenge coordinated as part of MICCAI 2017 and has been deployed onto Microsoft's Azure platform for use in measuring the progression of WMH in patients diagnosed with Alzheimer's and other neurological disease states. We'll present an overview of the preprocessing, model architecture, training, post-processing, and deployment of the WMH model. 25 Minute Talk Vivek Gupta - Medical Doctor, Mayo Clinic
Matt Berseth - Lead Scientist & Co-Founder, NLP Logix
Matt Berseth - Lead Scientist & Co-Founder, NLP Logix
Add to My Interests
DC7252 - The NVIDIA AI City Challenge - AI for Smarter Traffic and Safer Streets We'll introduce the of the AI cities challenge winners announced at GTC2017. Honghui Shi from University of Illinois at Urbana-Champaign who will do a ten minute presentation on multiple-Kernel based vehicle tracking Using 3D deformable models. Zheng Tang will then present on effective object detection from traffic camera videos. 25 Minute Talk Honghui Shi - Ph.D. Candidate, Image Formation and Processing group (IFP), UIUC, University of Illinois at Urbana-Champaign
Zheng Tang - Ph.D. Candidate , University of Washington
Milind Naphade - Chief Technology Officer, AI Cities, NVIDIA
Zheng Tang - Ph.D. Candidate, University of Washington
Add to My Interests
DC7245 - The Practical Application of AI to the Business of Law Watson legal conducted over 100 use case identification workshops to determine where we can and cannot add value by wrapping AI around our clients' business-of-law goals; learn about the top use cases as well as the metrics we used to evaluate and prioritize them. IBM Watson is the most implemented AI platform for business use in the world across industries; learn where and why your clients are investing in AI across their businesses so you can anticipate how related areas of law will be impacted. We are grossly unprepared for the impact of AI on the legal profession; what can we rationally expect the next 5 years to look like? 50 Minute Talk Brian Kuhn - Global Leader and Co-Creator of IBM Watson Legal, Artificial Intellige, IBM Watson
Add to My Interests
DC7230 - Understanding Neutrino Interactions Using Deep Learning The 2015 Nobel Prize in Physics was awarded for the discovery of neutrino oscillations, which indicates that neutrinos have mass. This phenomenon was unexpected and is one of the clearest signs of new physics beyond the Standard Model. The NOvA experiment aims to deepen our understanding of neutrino oscillations by measuring the properties of a muon neutrino beam produced at Fermi National Accelerator Laboratory at a Near Detector close to the beam source, and measuring the rate that muon neutrinos oscillate into electron neutrinos over an 810 km trip to a 14,000 ton Far Detector in Ash River, MN. Understanding this process may explain why the universe is made of matter instead of antimatter. Performing this measurement requires a high-precision method for classifying neutrino interactions. To this end, we developed a convolutional neural network that gave a 30 percent improvement in electron neutrino selection over previous methods – equivalent increasing the Far Detector mass by 4,000 tons. 25 Minute Talk Adam Aurisano - Assistant Professor, University of Cincinnati
Add to My Interests
DC7174 - Using AI to Read Human Body Language in Real-Time from Standard Video Inexpensive connected cameras in vehicles, drones, and buildings provide immense volumes of raw video imagery. It's practically impossible for humans to monitor and understand all of this footage to determine actionable events. We'll present a computer vision AI technology that uses deep learning to understand and read human body language from standard 2D RGB video cameras. We'll describe in detail the stages of our NVIDIA CUDA-based pipeline, from training on DGX-1 supercomputers to TITAN X cloud-solutions to edge-based deployment on Jetson TX2 modules. We'll describe how this system can be integrated into a variety of industrial applications, including human behavior monitoring for analytics and security; fall/accident detection in the home; and full body VR for collaboration and simulation. We'll also present live demos using a standard webcam and a Jetson TX2 developer kit. 25 Minute Talk Paul Kruszewski - CEO, wrnch
Add to My Interests
DC7178 - Using CNTK Train Reinforcement Learning Model Play Game with GPUs in Azure We're introducing Microsoft's Cognitive Toolkit (CNTK) and GPU in Azure, using both to train an AI model. We trained a neural network to play Hangman by appropriately guessing letters of a partially or fully obscured word. The network receives an input of a representation of the word (total number of characters, the identity of any revealed letters), and a list of which letters have been guessed so far. Then, the network guesses the next letter. This repo shows our method for training the network with CNTK and validating its performance on a withheld test set, as well as operationalizing the model for gameplay on an Azure Web App. 50 Minute Talk Vicky Fu - Cloud Solution Architect, Microsoft, Microsoft
Add to My Interests
DC7204 - Video Transcoding and Machine Learning at the Tactical Edge In this talk, we will provide an overview of GPU based video dissemination architecture useful for distributing video across disadvantaged networks in a compute-constrained environment. This architecture is well suited for enabling machine learning applications at the tactical edge. We will discuss the challenges that exist at the intersection of machine learning and video transcoding, specifically with regards to real-time object detection. 25 Minute Talk Vu Tran - Senior Data Scientist, Booz Allen Hamilton
Christopher Mucciolo - Software Engineer, Booz Allen Hamilton
Add to My Interests
DC7221 - Virtualization Infrastructure Decisions This session features senior-level technical decision makers from the government and architects who help them reach their virtualization infrastructure decisions. Questions will be backward-looking (how did you arrive at what you did and what would you do differently?), current state (maintaining the environment requires resources, give resource advice for those planning move from physical to virtual), and forward-looking (what can't you do that you would like/need to do?). 50 Minute Panel Chip Carr - Product Specialist, NVIDIA GRID Technology, NVIDIA
Branden Belush - Principal Systems Integration Engineer, Stratus IT
Bill Hackley - Senior Systems Engineer - Federal, Nutanix
Branden Belush - Principal Systems Integration Engineer , Stratus IT
David Weissman - Principal Sales Engineer US Special Programs, Citrix
Andrew Schilling - Chief Infrastructure Officer, Cannon Design
Chip Carr - Product Specialist, NVIDIA GRID Technology, NVIDIA
Add to My Interests
SEWGTC - Women@GTC Networking Breakfast

Join us for our first GTC DC women’s networking event. It’s an opportunity to engage with women doing important work in artificial intelligence, computer vision, and virtual reality.

Meet fellow attendees, speakers, and NVIDIA employees in a casual networking environment. Catch up with your colleagues or meet women who have just entered the field -- it’s a great way to build your network of artificial intelligence experts.

Please RSVP so we have breakfast for everyone.  

The event is open to any GTC attendees who support diversity and inclusion in the tech world. 

60 Minute Special Event
Add to My Interests
DC7213 - World's Fastest Machine Learning With GPUs Deep learning algorithms have benefited greatly from the recent performance gains of GPUs. However, it has been unclear whether GPUs can speed up machine learning algorithms such as generalized linear modeling, random forests, gradient boosting machines, and clustering. H2O.ai, the leading open source AI company, is bringing the best-of-breed data science and machine learning algorithms to GPUs. We'll introduce H2O4GPU, a fully featured machine learning library that is optimized for GPUs with a robust Python API that is a drop-dead replacement for scikit-learn. We'll demonstrate benchmarks for the most common algorithms relevant to enterprise AI and showcase performance gains as compared to running on CPUs. 25 Minute Talk Jon Mckinney - Senior Developer, H2O.ai
Add to My Interests
Remove From Schedule Add To Schedule Are you sure you would like to Delete this personal time? Edit My Schedule Edit Personal Time This session is full. Would you like to be added to the waiting list? Would you like to remove "{0}" from your schedule? Would you like to add "{0}" from your schedule? Sorry, this session is full. Waitlist Available Sorry, this session and it's waiting list are completely full. Sessions Available Adding this multi-day session automatically enrolls you for all times shown below. Removing this multi-day session automatically removes you for all times shown below. Adding this multi-day session automatically enrolls you for all session times for this session. Removing this multi-day session automatically removes you for all session times for this session. Click to view details Interests Hide Interests Search Sessions Export Schedule There is a scheduling conflict. You cannot add this session to your schedule because you are participating in another session at this time. Schedule Conflict. An error occurred while processing this request.. Adding this item creates a conflict with another session on your schedule. Remove from Waiting List Add to waiting list Removing this will remove you from the waiting list for all session times for this session Adding this will add you to the waiting list for all session times for this session. You have nothing scheduled Tap below to see a list of sessions and activities that are available to add to your schedule this week Choose from the list of sessions to the left to add to your schedule for the day Add a Session
Get More Results