对比学习中表征损害与分配损害的研究
An Investigation of Representation and Allocation Harms in Contrastive Learning
AI总结：
本研究针对自监督学习中少数群体表征损害研究不足的问题，验证了对比学习会引发少数群体表征坍缩的现象，通过因果分析明确其与下游分配损害的关联，并给出理论解释。
摘要：
在监督学习场景中，样本占比不足对少数群体性能的影响是已知的严重问题，但目前在自监督学习（self-supervised learning, SSL）语境下相关研究仍不充分。本文证明，对比学习（contrastive learning, CL）这一主流SSL变体，会倾向于将少数群体的表征与特定多数群体的表征发生坍缩。我们将该现象定义为表征损害，并在图像和文本数据集上采用对应的主流CL方法验证了该现象。此外，我们针对下游分类任务中的分配损害开展了因果中介分析，结果显示表征损害是造成分配损害的部分原因，凸显了研究并缓解表征损害的重要性。最后，我们利用随机块模型为表征损害提供了理论解释，该模型会在对比学习场景下引发表征性神经坍缩。
英文摘要：
The effect of underrepresentation on the performance of minority groups is known to be a serious problem in supervised learning settings; however, it has been underexplored so far in the context of self-supervised learning (SSL). In this paper, we demonstrate that contrastive learning (CL), a popular variant of SSL, tends to collapse representations of minority groups with certain majority groups. We refer to this phenomenon as representation harm and demonstrate it on image and text datasets using the corresponding popular CL methods. Furthermore, our causal mediation analysis of allocation harm on a downstream classification task reveals that representation harm is partly responsible for it, thus emphasizing the importance of studying and mitigating representation harm. Finally, we provide a theoretical explanation for representation harm using a stochastic block model that leads to a representational neural collapse in a contrastive learning setting.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
