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 2203.02586cs.LGcs.CV
收藏

面向分布外检测器的基于概念的解释

Concept-based Explanations for Out-Of-Distribution Detectors

University of Wisconsin - Madison(威斯康星大学麦迪逊分校);University of Michigan(密歇根大学)

Jihye Choi, Jayaram Raghuram, Ryan Feng, Jiefeng Chen, Somesh Jha, Atul Prakash

AI总结:

针对 OOD 检测器决策难解释的问题,提出检测完备性与概念可分性指标及无监督概念学习框架,为多种现成检测器提供基于概念的解释和决策推理。

摘要:

分布外(OOD)检测在确保深度神经网络(DNN)分类器安全部署方面发挥着关键作用。尽管大量方法都聚焦于提升 OOD 检测器的性能,但在解释其决策方面仍存在一个关键空白。我们通过基于学习到的高层概念为 OOD 检测器提供解释,帮助弥合这一空白。我们首先提出两项新指标,用于评估特定概念集合在解释 OOD 检测器时的有效性:1)检测完备性,用于量化概念在解释 OOD 检测器决策方面的充分性;2)概念可分性,用于刻画概念空间中分布内数据与 OOD 数据之间的分布分离程度。基于这些指标,我们提出一个无监督框架,用于学习一组兼具高检测完备性和概念可分性等理想属性的概念,并展示其在为多种现成 OOD 检测器提供基于概念的解释方面的有效性。我们还展示了如何识别促成检测结果的显著概念,并对其决策提供进一步推理。

英文摘要:

Out-of-distribution (OOD) detection plays a crucial role in ensuring the safe deployment of deep neural network (DNN) classifiers. While a myriad of methods have focused on improving the performance of OOD detectors, a critical gap remains in interpreting their decisions. We help bridge this gap by providing explanations for OOD detectors based on learned high-level concepts. We first propose two new metrics for assessing the effectiveness of a particular set of concepts for explaining OOD detectors: 1) detection completeness, which quantifies the sufficiency of concepts for explaining an OOD-detector's decisions, and 2) concept separability, which captures the distributional separation between in-distribution and OOD data in the concept space. Based on these metrics, we propose an unsupervised framework for learning a set of concepts that satisfy the desired properties of high detection completeness and concept separability, and demonstrate its effectiveness in providing concept-based explanations for diverse off-the-shelf OOD detectors. We also show how to identify prominent concepts contributing to the detection results, and provide further reasoning about their decisions.

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。

↑