The Decoupling Concept Bottleneck Model

Abstract

The Concept Bottleneck Model (CBM) is an interpretable neural network that leverages high-level concepts to explainmodel decisions and conduct human-machine interaction. However, in real-world scenarios, the deficiency of informative concepts canimpede the model’s interpretability and subsequent interventions. This paper proves that insufficient concept information can lead to aninherent dilemma of concept and label distortions in CBM. To address this challenge, we propose the Decoupling Concept BottleneckModel (DCBM), which comprises two phases: 1) DCBM for prediction and interpretation, which decouples heterogeneous informationinto explicit and implicit concepts while maintaining high label and concept accuracy, and 2) DCBM for human-machine interaction,which automatically corrects labels and traces wrong concepts via mutual information estimation. The construction of the interactionsystem can be formulated as a light min-max optimization problem. Extensive experiments expose the success of alleviatingconcept/label distortions, especially when concepts are insufficient. In particular, we propose the Concept Contribution Score (CCS) toquantify the interpretability of DCBM. Numerical results demonstrate that CCS can be guaranteed by the Jensen-Shannon divergenceconstraint in DCBM. Moreover, DCBM expresses two effective human-machine interactions, including forward intervention andbackward rectification, to further promote concept/label accuracy via interaction with human experts.

Tasks

Reproductions