arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2104.12437cs.LGstat.ML
收藏

走向严格解释:特征归属的形式化

Towards Rigorous Interpretations: a Formalisation of Feature Attribution

Deezer Research(迪泽研究);LIP6(LIP6实验室)

Darius Afchar, Romain Hennequin, Vincent Guigue

摘要:

特征归因通常被松散地表示为选择相关特征的子集作为预测的基本原理的过程。然而,由于任务的性质,文献中遇到的“相关性”的精确定义并不总是一致的。这种缺乏明确性的原因是,我们通常无法获得任何关于地面真相归属的概念,以及关于什么是好的解释的更一般性的辩论。在本文中,我们提出了正式的特征选择/属性的概念放松的功能依赖的基础上。特别是,我们将我们的概念扩展到实例设置,并推导出候选选择解决方案的必要属性,同时为任务依赖性留下空间。通过在合成数据集上计算地面实况属性,我们评估了许多最先进的属性方法,并表明即使在优化时,有些方法也无法验证所提出的属性并提供错误的解决方案。

英文摘要:

Feature attribution is often loosely presented as the process of selecting a subset of relevant features as a rationale of a prediction. Task-dependent by nature, precise definitions of "relevance" encountered in the literature are however not always consistent. This lack of clarity stems from the fact that we usually do not have access to any notion of ground-truth attribution and from a more general debate on what good interpretations are. In this paper we propose to formalise feature selection/attribution based on the concept of relaxed functional dependence. In particular, we extend our notions to the instance-wise setting and derive necessary properties for candidate selection solutions, while leaving room for task-dependence. By computing ground-truth attributions on synthetic datasets, we evaluate many state-of-the-art attribution methods and show that, even when optimised, some fail to verify the proposed properties and provide wrong solutions.

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。

↑