Convex SGD: Generalization Without Early Stopping

arXiv:2401.04067v2 Announce Type: replace
Abstract: We consider the generalization error associated with stochastic gradient descent on a smooth convex function over a compact set. We show the first bound on the generalization error that vanishes when the number of iterations $T$ and the dataset size $n$ go to zero at arbitrary rates; our bound scales as $\tilde{O}(1/\sqrt{T} + 1/\sqrt{n})$ with step-size $\alpha_t = 1/\sqrt{t}$. In particular, strong convexity is not needed for stochastic gradient descent to generalize well.

abstract arxiv compact cs.lg dataset error function go to gradient math.st set show stat.th stochastic type