Stochastic Approximation Approaches to Group Distributionally Robust Optimization

by   Moshe Y. Vardi, et al.

This paper investigates group distributionally robust optimization (GDRO), with the purpose to learn a model that performs well over m different distributions. First, we formulate GDRO as a stochastic convex-concave saddle-point problem, and demonstrate that stochastic mirror descent (SMD), using m samples in each iteration, achieves an O(m (log m)/ϵ^2) sample complexity for finding an ϵ-optimal solution, which matches the Ω(m/ϵ^2) lower bound up to a logarithmic factor. Then, we make use of techniques from online learning to reduce the number of samples required in each round from m to 1, keeping the same sample complexity. Specifically, we cast GDRO as a two-players game where one player simply performs SMD and the other executes an online algorithm for non-oblivious multi-armed bandits. Next, we consider a more practical scenario where the number of samples that can be drawn from each distribution is different, and propose a novel formulation of weighted DRO, which allows us to derive distribution-dependent convergence rates. Denote by n_i the sample budget for the i-th distribution, and assume n_1 ≥ n_2 ≥⋯≥ n_m. In the first approach, we incorporate non-uniform sampling into SMD such that the sample budget is satisfied in expectation, and prove the excess risk of the i-th distribution decreases at an O(√(n_1 log m)/n_i) rate. In the second approach, we use mini-batches to meet the budget exactly and also reduce the variance in stochastic gradients, and then leverage stochastic mirror-prox algorithm, which can exploit small variances, to optimize a carefully designed weighted DRO problem. Under appropriate conditions, it attains an O((log m)/√(n_i)) convergence rate, which almost matches the optimal O(√(1/n_i)) rate of only learning from the i-th distribution with n_i samples.


page 1

page 2

page 3

page 4


On-Demand Sampling: Learning Optimally from Multiple Distributions

Social and real-world considerations such as robustness, fairness, socia...

Stochastic Canonical Correlation Analysis

We tightly analyze the sample complexity of CCA, provide a learning algo...

Stochastic Multi-level Composition Optimization Algorithms with Level-Independent Convergence Rates

In this paper, we study smooth stochastic multi-level composition optimi...

Optimal Spend Rate Estimation and Pacing for Ad Campaigns with Budgets

Online ad platforms offer budget management tools for advertisers that a...

Stochastic Optimization with Bandit Sampling

Many stochastic optimization algorithms work by estimating the gradient ...

Sample Complexity of Asynchronous Q-Learning: Sharper Analysis and Variance Reduction

Asynchronous Q-learning aims to learn the optimal action-value function ...

Online Estimation and Optimization of Utility-Based Shortfall Risk

Utility-Based Shortfall Risk (UBSR) is a risk metric that is increasingl...

Please sign up or login with your details

Forgot password? Click here to reset