Self-paced Data Augmentation for Training Neural Networks

TL;DR

This paper introduces self-paced augmentation (SPA), a dynamic method for selecting suitable samples for data augmentation in neural network training, improving generalization especially with limited data.

Contribution

The paper proposes SPA, a novel approach that automatically and adaptively selects samples for data augmentation, outperforming existing methods like RandAugment.

Findings

SPA improves generalization performance with small datasets

SPA outperforms RandAugment in experiments

Effective sample selection enhances data augmentation benefits

Abstract

Data augmentation is widely used for machine learning; however, an effective method to apply data augmentation has not been established even though it includes several factors that should be tuned carefully. One such factor is sample suitability, which involves selecting samples that are suitable for data augmentation. A typical method that applies data augmentation to all training samples disregards sample suitability, which may reduce classifier performance. To address this problem, we propose the self-paced augmentation (SPA) to automatically and dynamically select suitable samples for data augmentation when training a neural network. The proposed method mitigates the deterioration of generalization performance caused by ineffective data augmentation. We discuss two reasons the proposed SPA works relative to curriculum learning and desirable changes to loss function instability. Experimental results demonstrate that the proposed SPA can improve the generalization performance, particularly when the number of training samples is small. In addition, the proposed SPA outperforms the state-of-the-art RandAugment method.