arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2303.16992cs.CL
收藏

ContraSim——基于对比学习分析神经表示

ContraSim -- Analyzing Neural Representations Based on Contrastive Learning

Technion – Israel Institute of Technology(以色列理工学院)

Adir Rahamim, Yonatan Belinkov

AI总结:

针对现有相似性度量在标准基准上表现平庸的问题,提出基于对比学习的参数化度量ContraSim,在语言和视觉模型的标准及新基准上准确率显著提升,并揭示了神经网络分析的新见解。

摘要:

近期的研究通过基于相似性的分析来比较神经网络表示,以改进模型解释。相似性度量的质量通常通过其能否为预期匹配的表示赋予高分来评估。然而,现有的相似性度量在标准基准上表现平庸。在这项工作中,我们开发了一种基于对比学习的新相似性度量,称为ContraSim。与常见的闭式相似性度量不同,ContraSim通过使用相似和不相似的样本来学习参数化度量。我们对方法进行了广泛的实验评估,涵盖了语言和视觉模型,在标准层预测基准以及我们引入的两个新基准:多语言基准和图像-标题基准上进行测试。在所有情况下,ContraSim的准确率远高于以往的相似性度量,即使在面对具有挑战性的样本时也是如此。最后,ContraSim更适合用于神经网络分析,揭示了以往度量未捕获的新见解。

英文摘要:

Recent work has compared neural network representations via similarity-based analyses to improve model interpretation. The quality of a similarity measure is typically evaluated by its success in assigning a high score to representations that are expected to be matched. However, existing similarity measures perform mediocrely on standard benchmarks. In this work, we develop a new similarity measure, dubbed ContraSim, based on contrastive learning. In contrast to common closed-form similarity measures, ContraSim learns a parameterized measure by using both similar and dissimilar examples. We perform an extensive experimental evaluation of our method, with both language and vision models, on the standard layer prediction benchmark and two new benchmarks that we introduce: the multilingual benchmark and the image-caption benchmark. In all cases, ContraSim achieves much higher accuracy than previous similarity measures, even when presented with challenging examples. Finally, ContraSim is more suitable for the analysis of neural networks, revealing new insights not captured by previous measures.

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。

↑