On Feature Normalization and Data Augmentation

TL;DR

This paper introduces Moment Exchange, a feature space data augmentation technique that replaces feature moments between images to enhance recognition model generalization, inspired by their role in image generation.

Contribution

It proposes a novel implicit data augmentation method that leverages feature moments for recognition tasks, bridging the gap between their roles in image generation and recognition.

Findings

Improves recognition accuracy across multiple benchmark datasets.

Enhances model generalization with minimal computational overhead.

Compatible with existing augmentation methods.

Abstract

The moments (a.k.a., mean and standard deviation) of latent features are often removed as noise when training image recognition models, to increase stability and reduce training time. However, in the field of image generation, the moments play a much more central role. Studies have shown that the moments extracted from instance normalization and positional normalization can roughly capture style and shape information of an image. Instead of being discarded, these moments are instrumental to the generation process. In this paper we propose Moment Exchange, an implicit data augmentation method that encourages the model to utilize the moment information also for recognition models. Specifically, we replace the moments of the learned features of one training image by those of another, and also interpolate the target labels -- forcing the model to extract training signal from the moments in addition to the normalized features. As our approach is fast, operates entirely in feature space, and mixes different signals than prior methods, one can effectively combine it with existing augmentation approaches. We demonstrate its efficacy across several recognition benchmark data sets where it improves the generalization capability of highly competitive baseline networks with remarkable consistency.