On Guaranteed Optimal Robust Explanations for NLP Models

TL;DR

This paper introduces a method for generating optimal, robust explanations for NLP models by selecting minimal word subsets that ensure consistent predictions despite embedding perturbations, improving interpretability and bias detection.

Contribution

It develops novel algorithms based on implicit hitting sets and universal subsets for computing robust explanations, enhancing existing NLP explanation frameworks with optimality and robustness guarantees.

Findings

Effective explanations for sentiment analysis tasks

Ability to detect bias through constrained explanations

Improved robustness of explanations against embedding perturbations

Abstract

We build on abduction-based explanations for ma-chine learning and develop a method for computing local explanations for neural network models in natural language processing (NLP). Our explanations comprise a subset of the words of the in-put text that satisfies two key features: optimality w.r.t. a user-defined cost function, such as the length of explanation, and robustness, in that they ensure prediction invariance for any bounded perturbation in the embedding space of the left out words. We present two solution algorithms, respectively based on implicit hitting sets and maximum universal subsets, introducing a number of algorithmic improvements to speed up convergence of hard instances. We show how our method can be con-figured with different perturbation sets in the em-bedded space and used to detect bias in predictions by enforcing include/exclude constraints on biased terms, as well as to enhance existing heuristic-based NLP explanation frameworks such as Anchors. We evaluate our framework on three widely used sentiment analysis tasks and texts of up to100words from SST, Twitter and IMDB datasets,demonstrating the effectiveness of the derived explanations.