Conference:  Transform X 2022

Authors: Anima Anandkumar

2022-10-19

As the Director of AI Research at NVIDIA, Dr. Anandkumar shares some of the highest-impact topics her team is researching—from better weather modeling to tackle climate change, to modeling CO2 carbon capture and storage, to coronavirus aerosol simulation and drug discovery. Previously, Anandkumar helped launch Amazon SageMaker, Comprehend, and Rekognition at AWS (Amazon Web Services). She is also the Bren Professor at Caltech’s CMS (Computing + Mathematical Sciences) Department and serves as part of the expert network of the World Economic Forum.Building on her seminal paper that covers the initial development of tensor algorithms, Dr. Anandkumar presents her research on Fourier Neural Operators (FNOs), which in some cases can replace computationally costly Navier-Stokes equations that underpin many fluid dynamics simulations such as weather forecasting models and drug discovery processes. She advocates the use of the Fourier transform in neural networks to make them discretization- (or even quantization-) invariant. By replacing brute-force computational approaches with FNOs running on GPUs, Dr. Anandkumar is able to reduce the complexity of the simulation of weather modeling (a 45,000x speed-up) and carbon capture and sequestration (a more than 10,000x speed-up).Join Dr. Anandkumar as she will inspire attendees to apply FNOs, smarter model architectures, and parallel computation to solve the public health and climate crises of our time.

Conference:  KubeCon + CloudNativeCon Europe 2022

Authors: Jose Navarro, Prayana Galih

2022-05-18

The adoption of MLOps practices and tooling by organizations has considerably reduced the pain points to productionise Machine Learning models. However, with the increase of the number of models available by a company to deploy, the diversity of frameworks used to train those models and the different infrastructure required to run each model, new challenges arise for Machine Learning Platform teams e.g: How can we deploy new models from the same or different frameworks concurrently? How can we improve throughput and optimize resource utilization in our serving infrastructure, especially GPUs? Cookpad ML Platform Engineers will talk in this session how Triton Inference Server, an open-source model serving tool from Nvidia, can simplify the process of model deployment and optimise the resource utilisation by efficiently supporting concurrent models on single GPU or CPU, and multi-GPU servers.Click here to view captioning/translation in the MeetingPlay platform!