Daguang Xu

Daguang Xu is now a research manager at AI-Infra of NVIDIA. He is leading a research team in healthcare AI, focusing on developing world-class machine learning and deep learning-based methods to solve the challenging problems in medical domain. His current research interest includes but not limited to medical imaging analysis, EHR analysis, computer aided diagnosis, deep learning, pattern recognition and computer vision, etc.

Personal webpage: http://daguangxu.net/

Andriy Myronenko

Andriy Myronenko is a senior research scientist at NVIDIA. His research interests include computer vision, deep learning and medical image analysis. At NVIDIA, his focus is on innovative AI algorithms for medical imaging applications, such as 3D MRI/CT organ and tumor segmentation. He also works on enabling AI technologies to be a part of clinical workflow  through collaborations with hospitals, including UCSF and Stanford. In 2018, he won the first place in the largest international challenge of 3D MRI brain tumor segmentation (BraTS).  

Holger Roth

Holger Roth, a Principal Federated Learning Scientist at NVIDIA, specializes in developing distributed and collaborative software and models for various industries using federated learning and analytics. He has been exploring the topic both from theoretical and practical standpoints. During the COVID-19 pandemic, he led the experimentation of a federated learning study involving twenty hospitals around the globe to train more generalizable models for predicting clinical outcomes in symptomatic patients.

Reference-Noise Compensation Scheme for Single- Ended Package-to-Package Links

We present a method for tracking and compensating the reference noise in single-ended links, enabling energy-efficient package-to-package communication in systems with noisy environment. The 25Gb/s/pin serial link with compensation loop is fabricated in 16nm FinFET process operating over PCB channel with >200mV reference noise. The maximum measured bandwidth of the compensation loop is 30MHz, and its energy and area penalty are negligible.

Yuval Atzmon

I work on generative vision models across different data modalities (like video, natural language, and action). My focus is on controllable and personalized image and video generation with fine-grained control over identity, motion, and compositional relationships.

I obtained my Ph.D. under the guidance of Prof. Gal Chechik at Bar-Ilan University. My M.Sc. and B.Sc. are in Electrical Engineering from the Technion, Israel Institute of Technology.

Neural Temporal Adaptive Sampling and Denoising

Despite recent advances in Monte Carlo path tracing at interactive rates, denoised image sequences generated with few samples per-pixel often yield temporally unstable results and loss of high-frequency details. We present a novel adaptive rendering method that increases temporal stability and image fidelity of low sample count path tracing by distributing samples via spatio-temporal joint optimization of sampling and denoising.