Stochastic Optimization for Spectral Risk Measures
Abstract
Spectral risk objectives – also called L-risks – allow for learning systems to interpolate between optimizing average-case performance (as in empirical risk minimization) and worst-case performance on a task. We develop LSVRG, a stochastic algorithm to optimize these quantities by characterizing their subdifferential and addressing challenges such as biasedness of subgradient estimates and non-smoothness of the objective. We show theoretically and experimentally that out-of-the-box approaches such as stochastic subgradient and dual averaging can be hindered by bias, whereas our approach exhibits linear convergence.
Cite
Text
Mehta et al. "Stochastic Optimization for Spectral Risk Measures." Artificial Intelligence and Statistics, 2023.Markdown
[Mehta et al. "Stochastic Optimization for Spectral Risk Measures." Artificial Intelligence and Statistics, 2023.](https://mlanthology.org/aistats/2023/mehta2023aistats-stochastic/)BibTeX
@inproceedings{mehta2023aistats-stochastic,
title = {{Stochastic Optimization for Spectral Risk Measures}},
author = {Mehta, Ronak and Roulet, Vincent and Pillutla, Krishna and Liu, Lang and Harchaoui, Zaid},
booktitle = {Artificial Intelligence and Statistics},
year = {2023},
pages = {10112-10159},
volume = {206},
url = {https://mlanthology.org/aistats/2023/mehta2023aistats-stochastic/}
}