Self-supervision of Feature Transformation for Further Improving Supervised Learning

TL;DR

This paper introduces a feature-based self-supervised pretext task for CNNs that reduces training overhead and enhances supervised learning by expanding labels with semantic information, leading to improved accuracy.

Contribution

The paper proposes a novel feature-based pretext task and a multi-classifier framework to incorporate self-supervision into CNN training with minimal overhead.

Findings

Improved accuracy on various supervised tasks

Effective use of feature transformations for self-supervision

Wide applicability across different CNN architectures

Abstract

Self-supervised learning, which benefits from automatically constructing labels through pre-designed pretext task, has recently been applied for strengthen supervised learning. Since previous self-supervised pretext tasks are based on input, they may incur huge additional training overhead. In this paper we find that features in CNNs can be also used for self-supervision. Thus we creatively design the \emph{feature-based pretext task} which requires only a small amount of additional training overhead. In our task we discard different particular regions of features, and then train the model to distinguish these different features. In order to fully apply our feature-based pretext task in supervised learning, we also propose a novel learning framework containing multi-classifiers for further improvement. Original labels will be expanded to joint labels via self-supervision of feature transformations. With more semantic information provided by our self-supervised tasks, this approach can train CNNs more effectively. Extensive experiments on various supervised learning tasks demonstrate the accuracy improvement and wide applicability of our method.