arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2408.08041cs.LGcs.AIstat.ML
收藏

无监督学习中的聪明汉斯效应

The Clever Hans Effect in Unsupervised Learning

Technische Universität Berlin(柏林工业大学);BIFOLD – Berlin Institute for the Foundations of Learning and Data(柏林学习与数据基础研究所);Aignostics(艾格诺斯蒂克斯公司);Freie Universität Berlin(柏林自由大学);Fraunhofer HHI(弗劳恩霍夫海因里希·赫兹研究所);Korea University(高丽大学);Max-Planck Institute for Informatics(马克斯·普朗克信息学研究所);Google Deepmind(谷歌DeepMind)

Jacob Kauffmann, Jonas Dippel, Lukas Ruff, Wojciech Samek, Klaus-Robert Müller, Grégoire Montavon

AI总结:

本文利用专门的可解释 AI 方法并结合理论分析,首次证明无监督学习中广泛存在聪明汉斯效应,指出归纳偏置是主要成因,并揭示相关应用风险与稳健化方向。

摘要:

无监督学习已成为 AI 系统的重要组成部分。它所产生的表示(例如在基础模型中)对于各种下游应用至关重要。因此,必须仔细检查无监督模型,不仅要确保它们产生准确的预测,还要确保这些预测并非“出于错误原因而正确”,即所谓的聪明汉斯(Clever Hans,CH)效应。利用专门开发的可解释 AI 技术,我们首次证明 CH 效应在无监督学习中广泛存在。我们的实证发现得到了理论洞见的补充;有趣的是,这些理论洞见指出,无监督学习机器中的归纳偏置是 CH 效应的主要来源。总体而言,我们的工作揭示了与无监督学习实际应用相关的、尚未被探索的风险,并提出了使无监督学习更稳健的方法。

英文摘要:

Unsupervised learning has become an essential building block of AI systems. The representations it produces, e.g. in foundation models, are critical to a wide variety of downstream applications. It is therefore important to carefully examine unsupervised models to ensure not only that they produce accurate predictions, but also that these predictions are not "right for the wrong reasons", the so-called Clever Hans (CH) effect. Using specially developed Explainable AI techniques, we show for the first time that CH effects are widespread in unsupervised learning. Our empirical findings are enriched by theoretical insights, which interestingly point to inductive biases in the unsupervised learning machine as a primary source of CH effects. Overall, our work sheds light on unexplored risks associated with practical applications of unsupervised learning and suggests ways to make unsupervised learning more robust.

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。

↑