Issues with post-hoc counterfactual explanations: a discussion

Abstract

Counterfactual post-hoc interpretability approaches have been proven to be useful tools to generate explanations for the predictions of a trained blackbox classifier. However, the assumptions they make about the data and the classifier make them unreliable in many contexts. In this paper, we discuss three desirable properties and approaches to quantify them: proximity, connectedness and stability. In addition, we illustrate that there is a risk for post-hoc counterfactual approaches to not satisfy these properties.

presented at 2019 ICML Workshop on Human in the Loop Learning (HILL 2019), Long Beach, USA

References in corpus (5)

• SmoothGrad: removing noise by adding noise
• How to Explain Individual Classification Decisions
• Towards Robust Interpretability with Self-Explaining Neural Networks
• Local Rule-Based Explanations of Black Box Decision Systems
• Interpretable Credit Application Predictions With Counterfactual Explanations

Cited by in corpus (8)

• Learning Model-Agnostic Counterfactual Explanations for Tabular Data
• A survey of algorithmic recourse: definitions, formulations, solutions, and prospects
• Keep Your Friends Close and Your Counterfactuals Closer: Improved Learning From Closest Rather Than Plausible Counterfactual Explanations in an Abstract Setting
• Achieving Diversity in Counterfactual Explanations: a Review and Discussion
• Interpretable Machine Learning: Moving From Mythos to Diagnostics
• Consistent Counterfactuals for Deep Models
• Flexible and Robust Counterfactual Explanations with Minimal Satisfiable Perturbations
• Towards User Guided Actionable Recourse