English

Towards learning to explain with concept bottleneck models: mitigating information leakage

Machine Learning 2022-11-08 v1 Cryptography and Security

Abstract

Concept bottleneck models perform classification by first predicting which of a list of human provided concepts are true about a datapoint. Then a downstream model uses these predicted concept labels to predict the target label. The predicted concepts act as a rationale for the target prediction. Model trust issues emerge in this paradigm when soft concept labels are used: it has previously been observed that extra information about the data distribution leaks into the concept predictions. In this work we show how Monte-Carlo Dropout can be used to attain soft concept predictions that do not contain leaked information.

Keywords

Cite

@article{arxiv.2211.03656,
  title  = {Towards learning to explain with concept bottleneck models: mitigating information leakage},
  author = {Joshua Lockhart and Nicolas Marchesotti and Daniele Magazzeni and Manuela Veloso},
  journal= {arXiv preprint arXiv:2211.03656},
  year   = {2022}
}
R2 v1 2026-06-28T05:20:41.017Z