Feature Augmentation for Self-supervised Contrastive Learning: A Closer Look
This work addresses the need for more diverse training data in self-supervised learning to enhance model generalization, but it is incremental as it builds on existing contrastive learning paradigms.
The paper tackles the problem of limited data diversity in self-supervised contrastive learning by proposing a feature augmentation framework to improve generalization and robustness. The result shows consistent performance gains in pre-trained feature learning, leading to better generalization in downstream tasks like image classification and object detection.
Self-supervised contrastive learning heavily relies on the view variance brought by data augmentation, so that it can learn a view-invariant pre-trained representation. Beyond increasing the view variance for contrast, this work focuses on improving the diversity of training data, to improve the generalization and robustness of the pre-trained models. To this end, we propose a unified framework to conduct data augmentation in the feature space, known as feature augmentation. This strategy is domain-agnostic, which augments similar features to the original ones and thus improves the data diversity. We perform a systematic investigation of various feature augmentation architectures, the gradient-flow skill, and the relationship between feature augmentation and traditional data augmentation. Our study reveals some practical principles for feature augmentation in self-contrastive learning. By integrating feature augmentation on the instance discrimination or the instance similarity paradigm, we consistently improve the performance of pre-trained feature learning and gain better generalization over the downstream image classification and object detection task.