A Resizable Mini-batch Gradient Descent based on a Multi-Armed Bandit

TL;DR

This paper introduces a resizable mini-batch gradient descent algorithm that adaptively selects batch sizes using a multi-armed bandit approach, outperforming fixed batch sizes and grid search in efficiency and effectiveness.

Contribution

The paper proposes a novel RMGD algorithm that dynamically adjusts batch sizes during training using multi-armed bandit strategies, reducing the need for exhaustive grid search.

Findings

RMGD outperforms fixed batch size methods in accuracy.

RMGD achieves better results than grid search in less time.

The adaptive approach effectively balances exploration and exploitation.

Abstract

Determining the appropriate batch size for mini-batch gradient descent is always time consuming as it often relies on grid search. This paper considers a resizable mini-batch gradient descent (RMGD) algorithm based on a multi-armed bandit for achieving best performance in grid search by selecting an appropriate batch size at each epoch with a probability defined as a function of its previous success/failure. This probability encourages exploration of different batch size and then later exploitation of batch size with history of success. At each epoch, the RMGD samples a batch size from its probability distribution, then uses the selected batch size for mini-batch gradient descent. After obtaining the validation loss at each epoch, the probability distribution is updated to incorporate the effectiveness of the sampled batch size. The RMGD essentially assists the learning process to explore the possible domain of the batch size and exploit successful batch size. Experimental results show that the RMGD achieves performance better than the best performing single batch size. Furthermore, it, obviously, attains this performance in a shorter amount of time than grid search. It is surprising that the RMGD achieves better performance than grid search.