Contextual Explanation Networks

Maruan Al-Shedivat, Avinava Dubey, Eric P. Xing

Abstract: We introduce contextual explanation networks (CENs)---a class of models that learn to predict by generating and leveraging intermediate explanations. CENs are deep networks that generate parameters for context-specific probabilistic graphical models which are further used for prediction and play the role of explanations. Contrary to the existing post-hoc model-explanation tools, CENs learn to predict and to explain jointly. Our approach offers two major advantages: (i) for each prediction, valid instance-specific explanations are generated with no computational overhead and (ii) prediction via explanation acts as a regularization and boosts performance in low-resource settings. We prove that local approximations to the decision boundary of our networks are consistent with the generated explanations. Our results on image and text classification and survival analysis tasks demonstrate that CENs are competitive with the state-of-the-art while offering additional insights behind each prediction, valuable for decision support.

TL;DR: A class of networks that generate simple models on the fly (called explanations) that act as a regularizer and enable consistent model diagnostics and interpretability.

Keywords: interpretability, regularization, deep learning, graphical models, model diagnostics, survival analysis

Code: [ alshedivat/cen](https://github.com/alshedivat/cen)

Data: [IMDb Movie Reviews](https://paperswithcode.com/dataset/imdb-movie-reviews)

Community Implementations: [ 4 code implementations](https://www.catalyzex.com/paper/arxiv:1705.10301/code)