Online Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 206 - Machine Learning Methodology
Type: Contributed
Date/Time: Tuesday, August 4, 2020 : 10:00 AM to 2:00 PM
Sponsor: Section on Statistical Learning and Data Science
Abstract #313599
Title: Adversarial Networks for Robust Estimation
Author(s): Ziyue Wang* and Zhiqiang Tan
Companies: Rutgers University-New Brunswick and Rutgers University
Keywords: robust estimation ; f-divergence; Huber's model; GAN

Consider robust estimation under Huber's contamination model. We develop and study a class of adversarial networks related to recent work on Generative Adversarial Networks to achieve computationally feasible robust estimation of locations and scatter matrices. The methods are constructed by minimizing f-divergence including the reverse KullbackÔÇôLeibler divergence and the total variation distance. We provide theory and numerical experiments to support the effectiveness of our methods.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2020 program