Dealing with the Evil Twins: Improving Random Augmentation by Addressing Catastrophic Forgetting of Diverse Augmentations

TL;DR

This paper enhances random data augmentation for out-of-distribution generalization by mitigating catastrophic forgetting caused by augmentation collisions, leading to improved performance across multiple benchmarks.

Contribution

It introduces a simple method to address catastrophic forgetting in random augmentation, significantly boosting its effectiveness for domain generalization.

Findings

Improved generalization on multiple benchmarks

Random augmentation performance enhanced by addressing forgetting

Mitigation of augmentation collisions improves feature learning

Abstract

Data augmentation is a promising tool for enhancing out-of-distribution generalization, where the key is to produce diverse, challenging variations of the source domain via costly targeted augmentations that maximize its generalization effect. Conversely, random augmentation is inexpensive but is deemed suboptimal due to its limited effect. In this paper, we revisit random augmentation and explore methods to address its shortcomings. We show that the stochastic nature of random augmentation can produce a set of colliding augmentations that distorts the learned features, similar to catastrophic forgetting. We propose a simple solution that improves the generalization effect of random augmentation by addressing forgetting, which displays strong generalization performance across various single source domain generalization (sDG) benchmarks.