arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2301.10367cs.LGcs.AI
收藏

迈向用于概念表征评估的鲁棒指标

Towards Robust Metrics for Concept Representation Evaluation

University of Cambridge(剑桥大学);Babylon Health(巴比伦健康公司);The Alan Turing Institute(阿兰·图灵研究所)

Mateo Espinosa Zarlenga, Pietro Barbiero, Zohreh Shams, Dmitry Kazhdan, Umang Bhatt, Adrian Weller, Mateja Jamnik

AI总结:

本文指出现有解耦纯度指标不适用于概念学习,提出评估概念表征纯度的新指标,并用于表征、干预和两类方法基准测试,发现单纯监督并不保证概念纯净。

摘要:

近期关于可解释性的工作集中在基于概念的解释上,即依据被称为概念的高级信息单元来解释深度学习模型。然而,研究表明概念学习模型容易在其表征中编码杂质,无法完全捕捉输入中有意义的特征。尽管概念学习缺乏衡量此类现象的指标,但解耦表示学习领域已经探索了数据中潜在变化因素这一相关概念,并有大量指标用于衡量这些因素的纯度。本文证明,此类指标并不适用于概念学习,并提出了用于评估两种方法中概念表征纯度的新型指标。我们展示了这些指标相对于现有指标的优势,并证明了它们在评估概念表征的鲁棒性以及对其执行的干预方面的效用。此外,我们展示了它们在基准测试来自两个方法家族的最先进方法时的效用,并发现与通常假设相反,仅靠监督可能不足以获得纯净的概念表征。

英文摘要:

Recent work on interpretability has focused on concept-based explanations, where deep learning models are explained in terms of high-level units of information, referred to as concepts. Concept learning models, however, have been shown to be prone to encoding impurities in their representations, failing to fully capture meaningful features of their inputs. While concept learning lacks metrics to measure such phenomena, the field of disentanglement learning has explored the related notion of underlying factors of variation in the data, with plenty of metrics to measure the purity of such factors. In this paper, we show that such metrics are not appropriate for concept learning and propose novel metrics for evaluating the purity of concept representations in both approaches. We show the advantage of these metrics over existing ones and demonstrate their utility in evaluating the robustness of concept representations and interventions performed on them. In addition, we show their utility for benchmarking state-of-the-art methods from both families and find that, contrary to common assumptions, supervision alone may not be sufficient for pure concept representations.

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。

↑