Visual correspondence-based explanations improve AI robustness and human-AI team accuracy
This work addresses the need for better AI explanations in high-stakes applications where humans make decisions, showing incremental improvements in robustness and team accuracy.
The paper tackles the problem of improving AI robustness and human-AI team accuracy by proposing self-interpretable image classifiers that use visual correspondences between query images and exemplars to explain and then predict. The result includes consistent improvements of 1 to 4 points on out-of-distribution datasets, with explanations found to be more useful to users and enabling complementary human-AI team accuracy higher than either AI-alone or human-alone in ImageNet and CUB tasks.
Explaining artificial intelligence (AI) predictions is increasingly important and even imperative in many high-stakes applications where humans are the ultimate decision-makers. In this work, we propose two novel architectures of self-interpretable image classifiers that first explain, and then predict (as opposed to post-hoc explanations) by harnessing the visual correspondences between a query image and exemplars. Our models consistently improve (by 1 to 4 points) on out-of-distribution (OOD) datasets while performing marginally worse (by 1 to 2 points) on in-distribution tests than ResNet-50 and a $k$-nearest neighbor classifier (kNN). Via a large-scale, human study on ImageNet and CUB, our correspondence-based explanations are found to be more useful to users than kNN explanations. Our explanations help users more accurately reject AI's wrong decisions than all other tested methods. Interestingly, for the first time, we show that it is possible to achieve complementary human-AI team accuracy (i.e., that is higher than either AI-alone or human-alone), in ImageNet and CUB image classification tasks.