Visual correspondence-based explanations improve AI robustness and human-AI team accuracy

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 (+1 to +4 points) on out-of-distribution (OOD) datasets while performing marginally worse (-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. * Equal contribution. Listing order is random. GN led the development of EMD-Corr and human studies on Gorilla. MRT led the development of CHM-Corr, pilot studies on HuggingFace, and the analysis of human-study data from Gorilla. AN advised the project. MRT's work was done before he joined University of Alberta. 36th Conference on Neural Information Processing Systems (NeurIPS 2022).