CVMay 25, 2022

Contrastive Learning with Boosted Memorization

arXiv:2205.12693v634 citationsh-index: 43Has Code
Originality Highly original
AI Analysis

This addresses the problem of learning representations from unlabeled, long-tailed data for computer vision and NLP applications, offering a novel data perspective approach.

The paper tackles self-supervised learning on long-tailed datasets by proposing Boosted Contrastive Learning (BCL), which uses the memorization effect of neural networks to enhance sample view discrepancies, achieving state-of-the-art results on benchmark datasets.

Self-supervised learning has achieved a great success in the representation learning of visual and textual data. However, the current methods are mainly validated on the well-curated datasets, which do not exhibit the real-world long-tailed distribution. Recent attempts to consider self-supervised long-tailed learning are made by rebalancing in the loss perspective or the model perspective, resembling the paradigms in the supervised long-tailed learning. Nevertheless, without the aid of labels, these explorations have not shown the expected significant promise due to the limitation in tail sample discovery or the heuristic structure design. Different from previous works, we explore this direction from an alternative perspective, i.e., the data perspective, and propose a novel Boosted Contrastive Learning (BCL) method. Specifically, BCL leverages the memorization effect of deep neural networks to automatically drive the information discrepancy of the sample views in contrastive learning, which is more efficient to enhance the long-tailed learning in the label-unaware context. Extensive experiments on a range of benchmark datasets demonstrate the effectiveness of BCL over several state-of-the-art methods. Our code is available at https://github.com/MediaBrain-SJTU/BCL.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes