Our manuscript on interpretable AI is on arXiv! In this paper, we demonstrate that deep CNNs for image classification can be modeled (in terms of necessary and sufficient conditions) to rationally inattentive
utility maximizers, a generative model used extensively in information economics for human decision making. We show that our interpretable model can mimic high-level classification performance of deep CNNs with high accuracy. To the best of our knowledge, this work is the first attempt at a ‘global’ approximation model for a neural network.