Interpreting vision and language generative models with semantic visual priors

When applied to Image-to-text models, explainability methods have two challenges. First, they often provide token-by-token explanations namely, they compute a visual explanation for each token of the generated sequence. This makes explanations expensive to compute and unable to comprehensively explain the model's output. Second, for models with visual inputs, explainability methods such as SHAP typically consider superpixels as features. Since superpixels do not correspond to semantically meaningful regions of an image, this makes explanations harder to interpret. We develop a framework based on SHAP, that allows for generating comprehensive, meaningful explanations leveraging the meaning representation of the output sequence as a whole. Moreover, by exploiting semantic priors in the visual backbone, we extract an arbitrary number of features that allows the efficient computation of Shapley values on large-scale models, generating at the same time highly meaningful visual explanations. We demonstrate that our method generates semantically more expressive explanations than traditional methods at a lower compute cost and that it can be generalized to a large family of vision-language models.