Diverse and Amortised Counterfactual Explanations for Uncertainty Estimates

To interpret uncertainty estimates, we extend recent work that generates multiple Counterfactual Latent Uncertainty Explanations (𝛿-CLUE), by applying additional constraints for diversity in the optimisation objective (∇-CLUE). We then propose a distinct method for discovering GLobal AMortised CLUEs (GLAM-CLUE) which learns mappings of arbitrary complexity between groups of uncertain and certain groups in a computationally efficient manner.

Appeared as a workshop paper at ICML 2021 (Algorithmic Recourse | Explainable AI | Socially Responsible ML | Uncertainty in Deep Learning).