NVDIA: Federated XGBoost - Formulation to Implementation


Apr 11, 12:00PM PDT(07:00PM GMT). Add to calendar: Google | Yahoo
  • Virtual SF Big Analytics
  • Free 9 Attendees
Description
Speaker

We are very excited to have invited two machine learning experts to discuss Federated XGBoost from NVIDIA. They have been directly contributing to the DMLC/xgboost repo as well as NVFLARE repo.

This talk will cover the XGBoost algorithm under a federated learning setting. We will discuss how federated XGBoost can be formulated and implemented. Specifically, what’s the distinction between traditional machine learning algorithms like XGBoost as compared with deep learning methods, and how it can impact the decision when we design and implement them. We will show practical examples under NVFlare..

Ziyue Xu(Nvidia)

Ziyue Xu
Dr. Ziyue Xu, IEEE Senior Member, is a Scientist at Nvidia. His research interests lie in the area of medical image analysis and machine learning.
Rong Ou
principal engineer at NVIDIA, currently working on enabling federated learning in XGBoost. Previously he worked on making the RAPIDS accelerator for Apache Spark more concurrent, adding out-of-core GPU support to XGBoost, and the Kubeflow MPI Operator. Before joining NVIDIA, he was a staff software engineer at Google..
*By RSVP, you submit information to event hosts, who will email you regarding events and services. You can opt-out from the emails at any time. Privacy Policy/GDPR.

Contact Organizer