arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2210.09404cs.LGcs.ITmath.IT
收藏

信息度量反映记忆模式

Measures of Information Reflect Memorization Patterns

Delhi Technological University(德里理工大学);Google Research India(谷歌研究院印度分部);Amazon Web Services(亚马逊网络服务);Technion – Israel Institute of Technology(以色列理工学院);Carnegie Mellon University(卡内基梅隆大学)

Rachit Bansal, Danish Pruthi, Yonatan Belinkov

AI总结:

本研究提出神经元激活模式多样性反映模型泛化与记忆的假设,通过信息论度量量化激活多样性,在多模态任务实验中验证该假设,发现信息组织可识别两类记忆,且成果可应用于模型选择。

摘要:

众所周知,神经网络会利用与目标标签共同出现的虚假伪影(或捷径),表现出启发式记忆。另一方面,已有研究表明网络会记忆训练样本,导致样本级记忆。这两类记忆都会阻碍网络在训练分布之外的泛化能力。检测这类记忆可能颇具挑战,通常需要研究人员精心定制测试集。在本研究中,我们提出假设并随后证明:不同神经元激活模式的多样性能够反映模型的泛化能力与记忆情况。我们通过信息论度量来量化神经激活的多样性,并在涵盖多个自然语言与视觉任务的实验中为我们的假设找到了支撑。重要的是,我们发现信息组织方式可指向这两种记忆形式,即便针对的是在无标签分布内样本上计算得到的神经激活亦是如此。最后,我们证明了本研究发现在模型选择问题中的实用性。相关代码与其他资源可访问https://rachitbansal.github.io/information-measures获取。

英文摘要:

Neural networks are known to exploit spurious artifacts (or shortcuts) that co-occur with a target label, exhibiting heuristic memorization. On the other hand, networks have been shown to memorize training examples, resulting in example-level memorization. These kinds of memorization impede generalization of networks beyond their training distributions. Detecting such memorization could be challenging, often requiring researchers to curate tailored test sets. In this work, we hypothesize -- and subsequently show -- that the diversity in the activation patterns of different neurons is reflective of model generalization and memorization. We quantify the diversity in the neural activations through information-theoretic measures and find support for our hypothesis on experiments spanning several natural language and vision tasks. Importantly, we discover that information organization points to the two forms of memorization, even for neural activations computed on unlabelled in-distribution examples. Lastly, we demonstrate the utility of our findings for the problem of model selection. The associated code and other resources for this work are available at https://rachitbansal.github.io/information-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。

↑