Dynamic Sparse Training versus Dense Training: The Unexpected Winner in Image Corruption Robustness

TL;DR

This paper demonstrates that Dynamic Sparse Training methods can outperform Dense Training in image corruption robustness without additional resource costs, challenging conventional beliefs in neural network training practices.

Contribution

It reveals that Dynamic Sparse Training can surpass Dense Training in robustness, especially at moderate sparsity levels, which is a novel insight in the field.

Findings

Dynamic Sparse Training outperforms Dense Training in robustness accuracy.

The advantage holds across images and videos, and various architectures.

Sparsity levels between 10% and 50% are most effective.

Abstract

It is generally perceived that Dynamic Sparse Training opens the door to a new era of scalability and efficiency for artificial neural networks at, perhaps, some costs in accuracy performance for the classification task. At the same time, Dense Training is widely accepted as being the "de facto" approach to train artificial neural networks if one would like to maximize their robustness against image corruption. In this paper, we question this general practice. Consequently, we claim that, contrary to what is commonly thought, the Dynamic Sparse Training methods can consistently outperform Dense Training in terms of robustness accuracy, particularly if the efficiency aspect is not considered as a main objective (i.e., sparsity levels between 10% and up to 50%), without adding (or even reducing) resource cost. We validate our claim on two types of data, images and videos, using several traditional and modern deep learning architectures for computer vision and three widely studied Dynamic Sparse Training algorithms. Our findings reveal a new yet-unknown benefit of Dynamic Sparse Training and open new possibilities in improving deep learning robustness beyond the current state of the art.