Oxford Nanopore Technologies is speeding the discovery of pathogens with its MinION device, a portable, low-cost, real-time DNA and RNA sequencer. Connecting to MinION is their hand-held AI supercomputer, MinIT. Powered by NVIDIA AGX™, MinIT enables DNA and RNA sequencing by anyone, anywhere, eliminating shipping-to-lab time and reducing time to answers from months to hours.

Scientists at the Massachusetts General Hospital (MGH) and Brigham and Women’s Hospital (BWH) Center for Clinical Data Science launched a federated learning initiative with Partners HealthCare. Clinicians and researchers will have access to diverse datasets to develop better AI models that can be used at the point of care. At the same time, the initiative will leverage these individual contributions to create a global model without compromising patient privacy. Using NVIDIA DGX deep learning systems and NVIDIA Clara’s privacy-preserving federated learning feature, the MGH & BWH Center for Clinical Data Science can develop and deploy AI across institutions.

King’s College London has partnered with NVIDIA and OWKIN to create a federated learning and algorithm deployment platform as part of the UK Research and Innovation (UKRI)-funded London Medical Imaging & Artificial Intelligence Centre for Value-Based Healthcare (AI4VBH). The platform will connect four NHS partner hospitals to form a decentralized dataset while protecting patient privacy. This federated dataset will be available to Centre partners and will be used to deliver research, clinical, and operational improvements in a wide range of areas, from cancer and heart failure to dementia and stroke. The AI4VBH Centre utilizes NVIDIA DGX-2™ and DGX-1 supercomputers and the NVIDIA Clara™ toolkit for highly optimized and accelerated AI.

With over 100,000 different products in its 4,700 U.S. stores, the Walmart Labs data science team must predict demand for 500 million items-by-store combinations every week. By performing forecasting with the NVIDIA RAPIDS™ suite of open-source data science and machine learning libraries built on CUDA-X AI™ and NVIDIA GPUs, the Walmart team is able to engineer machine learning features 100X faster and train algorithms 20X faster.

To make in-store retail experiences as streamlined as online experiences, Tracxpoint created the Artificial Intelligence Cart (AIC). Trained on the NVIDIA DGX Station™, inferenced using NVIDIA® TensorRT®, and able to perform real-time video analytics with the Deepstream SDK on NVIDIA Jetson™ TX2, AIC can recognize 100,000 products in under a second. Customers simply need to place the products in their carts. AIC also delivers personalized offers from suppliers in real time, helps customers navigate the supermarket with ease, and accepts automatic, digital payments.

AiFi is currently pilot testing NanoStore, their 24/7, autonomous, checkout-free store, with retail giants and universities. NanoStores hold over 500 different products and use image recognition, powered by NVIDIA T4 Tensor Core GPUs, to capture merchandise choices and add those to the customer’s tab.

5G will deliver multiple computing capabilities, including gigabit speeds with latencies under 20 ms. This has led the Verizon Envrmnt team to deploy powerful NVIDIA GPUs to beef up Verizon’s high-performance computing operations and create a distributed data center. 5G will also enable devices to become thinner, lighter, and more battery efficient, opening the door to memory-intensive parallel processing that can power rendering, deep learning, and computer vision. To take advantage of these advancements, Verizon is embedding NVIDIA GPUs throughout its network—at its mega data centers, at the hundreds of smaller ones that those feed, and at the thousands of smaller cell sites supported by those.

China Mobile utilizes NVIDIA EGX-powered 5G networks for enriched education and emergency services. Its Chengdu Research Institute, leveraging NVIDIA CloudXR and VR, equips underfunded schools with quality educational tools over 5G. Additionally, drones connected via 5G offer early, remote wildfire detection in inaccessible areas, enabling quick analysis and response by sending real-time visuals and data to emergency headquarters.

2Hz, Inc., is bringing clarity to live calls with noise-suppression technology powered by NVIDIA T4 and V100 GPUs. 2Hz’s deep learning algorithms scale up to 20X more than CPUs, and by running TensorRT on GPUs, 2Hz meets the 12 millisecond (ms) latency requirement for real-time communications.

Seagate is transforming manufacturing quality assurance using cutting-edge deep learning technologies, deployed from edge to cloud. With models trained on NVIDIA Tensor Core GPUs and optimized for inference at the edge, these solutions automate defect inspection of semiconductor wafers, electrical components, and assemblies—dramatically improving inspection accuracy, product quality, and factory efficiency.

Musashi utilizes AI technology in the optical inspection of defects in gears, camshafts, and welds, which has traditionally been a complex and manual process. Using NVIDIA DGX Station and Tensor Core GPUs, they developed and trained their TensorFlow-based network algorithms to analyze images and identify defects. Musashi was able to then quickly build and deploy a trained inference algorithm on NVIDIA Jetson on the manufacturing floor, giving them the ability to evolve and adapt their manufacturing process as needed.

Spartanburg, South Carolina, is the home of BMW Group’s largest plant in the world, producing 450 thousand vehicles per year. To ensure a flawless delivery experience, BMW uses GPUs to power intelligent image analytics that automate full-vehicle inspection. Using NVIDIA EGX edge servers, they can process data from a camera array capturing 1,000 images per car in 70 seconds. The system checks quality with the highest precision and confirms hundreds of customization options to match each customer’s exact order specifications.