Generalization of ERM in Stochastic Convex Optimization: The Dimension Strikes Back

In stochastic convex optimization the goal is to minimize a convex function over a convex set where is some unknown distribution and each in the support of is convex over . The optimization is commonly based on i.i.d.~samples from . A standard approach to such problems is empirical risk minimization (ERM) that optimizes . Here we consider the question of how many samples are necessary for ERM to succeed and the closely related question of uniform convergence of to over . We demonstrate that in the standard setting of Lipschitz-bounded functions over a of bounded radius, ERM requires sample size that scales linearly with the dimension . This nearly matches standard upper bounds and improves on dependence proved for setting by Shalev-Shwartz et al. (2009). In stark contrast, these problems can be solved using dimension-independent number of samples for setting and dependence for setting using other approaches. We further show that our lower bound applies even if the functions in the support of are smooth and efficiently computable and even if an regularization term is added. Finally, we demonstrate that for a more general class of bounded-range (but not Lipschitz-bounded) stochastic convex programs an infinite gap appears already in dimension 2.
View on arXiv