Abstract
We propose dynamic sampled stochastic approximation (SA) methods for stochastic optimization with a heavy-tailed distribution (with finite 2nd moment). The objective is the sum of a smooth convex function with a convex regularizer. Typically, it is assumed an oracle with an upper bound $\sigma^2$ on its variance (OUBV). Differently, we assume an oracle with \emph{multiplicative noise} (OMN). This rarely addressed setup is more aggressive but realistic, where the variance may not be bounded. Our methods achieve optimal iteration complexity and (near) optimal oracle complexity. For the smooth convex class, we use an accelerated SA method a la FISTA which achieves, given tolerance $\epsilon>0$, the optimal iteration complexity of $\mathcal{O}(\epsilon^{-\frac{1}{2}})$ with a near-optimal oracle complexity of $\mathcal{O}(\epsilon^{-2})[\ln(\epsilon^{-\frac{1}{2}})]^2$. This improves upon Ghadimi and Lan [Math. Program., 156:59-99, 2016] where it is assumed an OUBV. For the strongly convex class, our method achieves optimal iteration complexity of $\mathcal{O}(\ln(\epsilon^{-1}))$ and optimal oracle complexity of $\mathcal{O}(\epsilon^{-1})$. This improves upon Byrd et al. [Math Program, 134:127-155, 2012] where it is assumed an OUBV. In terms of variance, our bounds are local: they depend on variances $\sigma(x^*)^2$ at solutions $x^*$ and the per unit distance multiplicative variance $\sigma^2_L$. For the smooth convex class, our bounds can be tuned so to depend solely on $\sigma(x^*)^2$. For the strongly convex class, the more well-conditioned the problem is or the larger the initial batch size is, the more our bounds approach those obtained if it was assumed an OUBV with $\sigma^2:=\sigma(x^*)^2$. This bound is exactly achieved if the condition number is estimated. In case of an OUBV, our bounds are thus much sharper since typically $\max\{\sigma(x^*)^2,\sigma_L^2\}\ll\sigma^2$.

This publication has 36 references indexed in Scilit: