Adaptive Soft Contrastive Learning
This work addresses a limitation in self-supervised learning for visual representation learning, offering an incremental improvement as a plug-in module.
The paper tackles the problem that contrastive learning treats all samples as independent, which contradicts natural groupings in visual data, by proposing Adaptive Soft Contrastive Learning (ASCL) to introduce soft inter-sample relations, achieving the best performance on several benchmarks.
Self-supervised learning has recently achieved great success in representation learning without human annotations. The dominant method -- that is contrastive learning, is generally based on instance discrimination tasks, i.e., individual samples are treated as independent categories. However, presuming all the samples are different contradicts the natural grouping of similar samples in common visual datasets, e.g., multiple views of the same dog. To bridge the gap, this paper proposes an adaptive method that introduces soft inter-sample relations, namely Adaptive Soft Contrastive Learning (ASCL). More specifically, ASCL transforms the original instance discrimination task into a multi-instance soft discrimination task, and adaptively introduces inter-sample relations. As an effective and concise plug-in module for existing self-supervised learning frameworks, ASCL achieves the best performance on several benchmarks in terms of both performance and efficiency. Code is available at https://github.com/MrChenFeng/ASCL_ICPR2022.