CHIME: Causal Human-in-the-Loop Model Explanations

Abstract

Explaining the behaviour of Artificial Intelligence models has become a necessity. Their opaqueness and fragility are not tolerable in high-stakes domains especially. Although consider able progress is being made in the field of Explainable Artificial Intelligence, scholars have demonstrated limits and flaws of existing approaches: explanations requiring furtherinter pretation, non-standardised explanatory format, and overall fragility. In light of this fragmentation, we turn to the field of philosophy of science to understand what constitutesa good explanation, that is, a generalisation that covers both the actual outcome and, possibly multiple, counterfactual outcomes. Inspired by this, we propose CHIME: a human-in the-loop, post-hoc approach focused on creating such explanations by establishing the causal features in the input. We first elicit people’s cognitive abilities to understand what partso f the input the model might be attending to. Then, through Causal Discovery we uncover the underlying causal graph relating the different concepts. Finally, with such a structure, we compute the causal effects different concepts have towards amodel’s outcome. We evaluate the Fidelity, Coherence, and Accuracy of the explanations obtained with CHIME with respect to two state-of-the-art Computer Vision models trainedon real-world image data sets. We found evidence that the explanations reflect the causal concepts tied to a model’s prediction, both in terms of causal strength and accuracy.