解决长尾识别中对比学习失效问题，提升模型泛化能力。
Aligned Contrastive Loss for Long-Tailed Recognition
- 通过分析梯度冲突，设计对齐对比学习算法
- 在多个数据集上达到新最好性能
- 适合需要处理样本不均衡的视觉识别场景
本文提出一种对齐对比学习（ACL）算法，以应对长尾识别问题。研究发现，尽管多视角训练能提升性能，但随着视角数量增加，对比学习并未持续增强模型泛化能力。通过监督对比学习的理论梯度分析，我们识别出正负样本对间存在梯度冲突，以及吸引与排斥梯度不平衡的问题。所提出的ACL算法旨在消除这些缺陷，在长尾CIFAR、ImageNet、Places和iNaturalist数据集上进行了验证。实验结果表明，ACL在多个基准上均取得新的最优性能。
原文摘要 · Abstract (English)
In this paper, we propose an Aligned Contrastive Learning (ACL) algorithm to address the long-tailed recognition problem. Our findings indicate that while multi-view training boosts the performance, contrastive learning does not consistently enhance model generalization as the number of views increases. Through theoretical gradient analysis of supervised contrastive learning (SCL), we identify gradient conflicts, and imbalanced attraction and repulsion gradients between positive and negative pairs as the underlying issues. Our ACL algorithm is designed to eliminate these problems and demonstrates strong performance across multiple benchmarks. We validate the effectiveness of ACL through experiments on long-tailed CIFAR, ImageNet, Places, and iNaturalist datasets. Results show that ACL achieves new state-of-the-art performance.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。