Designing Counterfactual Generators using Deep Model Inversion
This work addresses the need for interpretable AI explanations in scenarios where training data is unavailable, offering a novel method for generating realistic counterfactuals, though it is incremental as it builds upon existing deep inversion techniques.
The paper tackles the problem of generating counterfactual explanations for black-box deep classifiers without access to training data, proposing DISC, which improves upon existing deep inversion methods by incorporating stronger priors, a manifold consistency objective, and progressive optimization, resulting in visually meaningful explanations that effectively learn decision boundaries and are robust to test-time corruptions.
Explanation techniques that synthesize small, interpretable changes to a given image while producing desired changes in the model prediction have become popular for introspecting black-box models. Commonly referred to as counterfactuals, the synthesized explanations are required to contain discernible changes (for easy interpretability) while also being realistic (consistency to the data manifold). In this paper, we focus on the case where we have access only to the trained deep classifier and not the actual training data. While the problem of inverting deep models to synthesize images from the training distribution has been explored, our goal is to develop a deep inversion approach to generate counterfactual explanations for a given query image. Despite their effectiveness in conditional image synthesis, we show that existing deep inversion methods are insufficient for producing meaningful counterfactuals. We propose DISC (Deep Inversion for Synthesizing Counterfactuals) that improves upon deep inversion by utilizing (a) stronger image priors, (b) incorporating a novel manifold consistency objective and (c) adopting a progressive optimization strategy. We find that, in addition to producing visually meaningful explanations, the counterfactuals from DISC are effective at learning classifier decision boundaries and are robust to unknown test-time corruptions.