Regularized Sample Average Approximation for High-Dimensional Stochastic Optimization Under Low-Rankness

TL;DR

This paper introduces a regularized sample average approximation method for high-dimensional stochastic matrix optimization, leveraging low-rank penalties to significantly reduce sample complexity and improve scalability.

Contribution

It proposes a novel RSAA approach with low-rank regularization, achieving near-linear sample complexity dependence on problem dimension, surpassing traditional SAA methods.

Findings

RSAA has sample complexity nearly linear in matrix dimension p.

Regularization reduces the required sample size for high-dimensional problems.

Results extend to high-dimensional low-rank matrix recovery beyond linear models.

Abstract

This paper concerns a high-dimensional stochastic programming problem of minimizing a function of expected cost with a matrix argument. To this problem, one of the most widely applied solution paradigms is the sample average approximation (SAA), which uses the average cost over sampled scenarios as a surrogate to approximate the expected cost. Traditional SAA theories require the sample size to grow rapidly when the problem dimensionality increases. Indeed, for a problem of optimizing over a $p$-by-$p$ matrix, the sample complexity of the SAA is given by $\tilde O(1)\cdot \frac{p^2}{\epsilon^2}\cdot{polylog}(\frac{1}{\epsilon})$ to achieve an $\epsilon$-suboptimality gap, for some poly-logarithmic function ${polylog}(\,\cdot\,)$ and some quantity $\tilde O(1)$ independent of dimensionality $p$ and sample size $n$. In contrast, this paper considers a regularized SAA (RSAA) with a low-rankness-inducing penalty. We demonstrate that the sample complexity of RSAA is $\tilde O(1)\cdot \frac{p}{\epsilon^3}\cdot {polylog}(p,\,\frac{1}{\epsilon})$, which is almost linear in $p$ and thus indicates a substantially lower dependence on dimensionality. Therefore, RSAA can be more advantageous than SAA especially for larger scale and higher dimensional problems. Due to the close correspondence between stochastic programming and statistical learning, our results also indicate that high-dimensional low-rank matrix recovery is possible generally beyond a linear model, even if the common assumption of restricted strong convexity is completely absent.