Choosing the Sample with Lowest Loss makes SGD Robust

[edit]

Vatsal Shah, Xiaoxia Wu, Sujay Sanghavi ;
Proceedings of the Twenty Third International Conference on Artificial Intelligence and Statistics, PMLR 108:2120-2130, 2020.

Abstract

The presence of outliers can potentially significantly skew the parameters of machine learning models trained via stochastic gradient descent (SGD). In this paper we propose a simple variant of the simple SGD method: in each step, first choose a set of k samples, then from these choose the one with the smallest current loss, and do an SGD-like update with this chosen sample. Vanilla SGD corresponds to $k=1$, i.e. no choice; $k>=2$ represents a new algorithm that is however effectively minimizing a non-convex surrogate loss. Our main contribution is a theoretical analysis of the robustness properties of this idea for ML problems which are sums of convex losses; these are backed up with synthetic and small-scale neural network experiments.

Related Material