SEBOOST - Boosting Stochastic Learning Using Subspace Optimization Techniques

TL;DR

SEBOOST is a versatile technique that enhances stochastic optimization algorithms by incorporating subspace optimization steps, leading to improved performance and robustness in deep learning tasks without significant additional computational cost.

Contribution

It introduces a novel subspace optimization approach that can be added to any stochastic optimizer, inspired by SESOP, improving performance and hyper-parameter robustness.

Findings

Boosts performance of various stochastic algorithms

Increases robustness to hyper-parameter changes

Achieves promising results on deep learning tasks

Abstract

We present SEBOOST, a technique for boosting the performance of existing stochastic optimization methods. SEBOOST applies a secondary optimization process in the subspace spanned by the last steps and descent directions. The method was inspired by the SESOP optimization method for large-scale problems, and has been adapted for the stochastic learning framework. It can be applied on top of any existing optimization method with no need to tweak the internal algorithm. We show that the method is able to boost the performance of different algorithms, and make them more robust to changes in their hyper-parameters. As the boosting steps of SEBOOST are applied between large sets of descent steps, the additional subspace optimization hardly increases the overall computational burden. We introduce two hyper-parameters that control the balance between the baseline method and the secondary optimization process. The method was evaluated on several deep learning tasks, demonstrating promising results.