JSAI2023

Presentation information

General Session

General Session » GS-1 Fundamental AI, theory

[1G3-GS-1] Fundamental AI, theory: algorithm

Tue. Jun 6, 2023 1:00 PM - 2:40 PM Room G (A4)

座長:金 秀明(NTT) [オンライン]

2:20 PM - 2:40 PM

[1G3-GS-1-05] A learning algorithm for robustly minimizing mean-variance of the loss distribution

〇Matthew James Holland1 (1. Osaka University)

Keywords:Variance control

Under losses which are potentially heavy-tailed, we consider the task of minimizing sums of the loss mean and standard deviation, without trying to accurately estimate the variance. By modifying a technique for variance-free robust mean estimation to fit our problem setting, we derive a simple learning procedure which can be easily combined with standard gradient-based solvers to be used in traditional machine learning workflows. Empirically, we verify that our proposed approach, despite its simplicity, performs as well or better than even the best-performing candidates derived from alternative criteria such as CVaR or DRO risks on a variety of datasets.

Authentication for paper PDF access

A password is required to view paper PDFs. If you are a registered participant, please log on the site from Participant Log In.
You could view the PDF with entering the PDF viewing password bellow.

Password