A Game Theoretic Approach to Class-wise Selective Rationalization

Abstract

Selection of input features such as relevant pieces of text has become a common technique of highlighting how complex neural predictors operate. The selection can be optimized post-hoc for trained models or incorporated directly into the method itself (self-explaining). However, an overall selection does not properly capture the multi-faceted nature of useful rationales such as pros and cons for decisions. To this end, we propose a new game theoretic approach to class-dependent rationalization, where the method is specifically trained to highlight evidence supporting alternative conclusions. Each class involves three players set up competitively to find evidence for factual and counterfactual scenarios. We show theoretically in a simplified scenario how the game drives the solution towards meaningful class-dependent rationales. We evaluate the method in single- and multi-aspect sentiment classification tasks and demonstrate that the proposed method is able to identify both factual (justifying the ground truth label) and counterfactual (countering the ground truth label) rationales consistent with human rationalization. The code for our method is publicly available.

Accepted by Neural Information Processing Systems (NeurIPS 2019), Vancouver, Canada

Cited by in corpus (13)

• Rationalization for Explainable NLP: A Survey
• Invariant Rationalization
• Explaining and Improving Model Behavior with k Nearest Neighbor Representations
• Self-Explaining Structures Improve NLP Models
• Interpreting Deep Learning Models in Natural Language Processing: A Review
• Explaining Deep Neural Networks
• Adversarial Infidelity Learning for Model Interpretation
• Generative causal explanations of black-box classifiers
• Rationalizing Predictions by Adversarial Information Calibration
• Learning from the Best: Rationalizing Prediction by Adversarial Information Calibration
• Model Interpretability and Rationale Extraction by Input Mask Optimization
• Learning to Give Checkable Answers with Prover-Verifier Games
• Paragraph-level Rationale Extraction through Regularization: A case study on European Court of Human Rights Cases