NVIDIA FLARE DAY 2024

NVIDIA FLARE DAY is an event dedicated to showcasing the cutting-edge applications of Federated Learning across various industries. See the talks from this year's event below.

Introduction

Opening Remarks, NVIDIA FLARE Overview & Roadmap

Headshot of Ankit Patel

Ankit Patel

Senior Director for Software Development Kits, APIs, and Tools at NVIDIA

Headshot of Chester Chen

Chester Chen

Acting Product Manager and Senior Engineering Manager of NVIDIA FLARE

Watch the Talk:
Video Link
View Slides:
Slides Link

NVIDIA AV Team

Autonomous Driving Federated Learning

To enable the training of AV models with combined CN+US multimodal raw sensor data, we proposed the NV-patented Round Robin Federated Learning (RRFL) and integrated it with the NVFLARE framework.

Headshot of Hanson (Haisheng) Xu

Hanson (Haisheng) Xu

Autonomous Driving Algorithm Engineer at NVIDIA

Headshot of Allen (Yichun) Shen

Allen (Yichun) Shen

Senior Engineering Manager for AV Perception at NVIDIA

Watch the Talk:
Video Link

Roche

Unlocking patient-level data at scale with federated computing to drive collaborative research and advance science

We'll present our exploration outcomes of Federated Analysis and Federated Learning in Personalized Healthcare to advance multicenter RWE studies and accelerate scientific discoveries.

Headshot of Eric Boernert

Eric Boernert

Product Manager for Federated Open Science at Roche

Headshot of Jacek Chmiel

Jacek Chmiel

Technology Expert for Federated Open Science at Roche

Headshot of Fahime Sheikhzadeh

Fahime Sheikhzadeh

Principal Research Scientist at Roche

Watch the Talk:
Video Link
View Slides:
Slides Link

Royal Bank of Canada

Confidential Federated Learning in a multi-party environment for product recommendations

By combining data mesh workflows with NVIDIA FLARE, product recommendations can be traced back to the specific data used for training, enabling business stakeholders to verify the fitness for business and regulatory compliance.

Headshot of Iustina Vintila

Iustina Vintila

Innovation Architect at Royal Bank of Canada

Watch the Talk:
Video Link
View Slides:
Slides Link

Rhino Health

Federated Computing In The Real World: From Drug Discovery to Fraud Detection

In this talk we will discuss some of the challenges we encountered at Rhino Health when working on real-world federated computing projects, as well as some tips for overcoming these challenges, highlighting some projects and use cases that were executed on the Rhino Federated Computing Platform.

Headshot of Tal Tiano Einat

Tal Tiano Einat

Lead Software Engineer at Rhino Health

Headshot of Chris Laws

Chris Laws

COO at Rhino Health

Watch the Talk:
Video Link
View Slides:
Slides Link

Flower Lab

Running Flower Projects on NVFLARE

Join us for an exciting technical talk on the groundbreaking integration between two leading federated learning frameworks: Flower and NVFlare. In this talk, we will show how the integration works, how you can use it, and what's next for Flower-on-Flare.

Headshot of Daniel J. Beutel

Daniel J. Beutel

CEO and co-founder of Flower

Watch the Talk:
Video Link

Apheris

Federated data collaboration in BioPharma: leveraging NVFlare and BioNeMo

This talk will focus on how computational governance, with NVFlare at its heart, revolutionizes federated training and evaluation. We'll present a practical case study of a BioNeMo implementation on Apheris' product, highlighting effective strategies for governance, privacy, and security.

Headshot of Robin Röhm

Robin Röhm

CEO and co-founder of Apheris

Headshot of Ellie Dobson

Ellie Dobson

VP of Product at Apheris

Watch the Talk:
Video Link
View Slides:
Slides Link

University of Wisconsin

Federated Learning in Medical Imaging: Enhancing Data Privacy and Advancing Healthcare

In this talk, we will discuss the primary motivations for adopting federated learning in medical imaging, delve into successful examples and their impact on improving diagnostic accuracy and patient outcomes, and envision the transformative potential to revolutionize personalized medicine, cross-institutional collaborations, and development of robust AI models.

Headshot of Dr. John W. Garrett

Dr. John W. Garrett

Associate Professor in Departments of Radiology and Medical Physics

Watch the Talk:
Video Link
View Slides:
Slides Link

Oak Ridge National Lab

Towards a Federating Learning Platform for HPC User Facilities

An overview of the ongoing effort to integrate NVFlare into the Oak Ridge Leadership Computing Facilities (OLCF) infrastructure to enable domain agnostic Federated Learning campaigns for HPC.

Headshot of Ryan Prout

Ryan Prout

Group Leader for Software Services Development Group in NCCS

Headshot of Alan Longcoy

Alan Longcoy

Software Engineer for Software Services Development Group at ORNL

Watch the Talk:
Video Link

Deloitte

FedRAG – A Secure Approach to Scaling Retrieval Augmented Generation

In this presentation, we introduce FedRAG, a new privacy-preserving solution for scaling RAG across a decentralized network of data providers. We also demonstrate how to implement FedRAG using the NVIDIA FLARE SDK and cloud-based trusted execution environments.

Headshot of Mohammad Manzari

Mohammad Manzari

Manager of Privacy Enhancing Technologies R&D group at Deloitte Consulting

Watch the Talk:
Video Link

OPTIMA

Enabling Precision Oncology with Federated Learning: Insights from the OPTIMA Consortium

Lisa Schneider has taken on a pivotal role in implementing Federated Learning within the OPTIMA Consortium. OPTIMA (Optimal Treatment for Patients with Solid Tumours in Europe Through Artificial intelligence) aims to advance treatments and improve decision-making processes for physicians and patients with prostate, breast and lung cancer using machine learning.

Headshot of Lisa Schneider

Lisa Schneider

Machine Learning Research Scientist at Bayer AG

FLARE Getting Started and LLM Support

End-to-end Pythonic APIs, LLM support - PEFT, SFT, RAG

Headshot of Holger Roth

Holger Roth

Principal Federated Learning Scientist at NVIDIA

Watch the Talk:
Video Link
View Slides:
Slides Link

Secure XGBoost

Using Homomorphic Encryption for Secure Federated XGBoost Learning

Headshot of Ziyue Xu

Ziyue Xu

Senior Scientist at NVIDIA

Watch the Talk:
Video Link
View Slides:
Slides Link

Confidential Computing and Looking Forward

How FLARE integrates with Confidential Computing, Closing Remarks

Headshot of Isaac Yang

Isaac Yang

Senior Software Engineer at NVIDIA

Headshot of Yan Cheng

Yan Cheng

Director of Engineering at NVIDIA and Chief Architect of NVIDIA FLARE

Watch the Talk:
Video Link
View Slides:
Slides Link