Assessing Robustness of Text Classification through Maximal Safe Radius Computation

TL;DR

This paper introduces a method to estimate the robustness of text classification models against word substitutions by approximating the maximal safe radius, providing guarantees on prediction stability.

Contribution

It proposes a novel framework combining Monte Carlo Tree Search and linear bounding techniques to approximate the maximal safe radius for NLP models, enhancing robustness analysis.

Findings

The methods effectively estimate robustness bounds across multiple datasets.

Robustness trends vary with different embeddings and models.

The framework offers insights for interpretability and model reliability.

Abstract

Neural network NLP models are vulnerable to small modifications of the input that maintain the original meaning but result in a different prediction. In this paper, we focus on robustness of text classification against word substitutions, aiming to provide guarantees that the model prediction does not change if a word is replaced with a plausible alternative, such as a synonym. As a measure of robustness, we adopt the notion of the maximal safe radius for a given input text, which is the minimum distance in the embedding space to the decision boundary. Since computing the exact maximal safe radius is not feasible in practice, we instead approximate it by computing a lower and upper bound. For the upper bound computation, we employ Monte Carlo Tree Search in conjunction with syntactic filtering to analyse the effect of single and multiple word substitutions. The lower bound computation is achieved through an adaptation of the linear bounding techniques implemented in tools CNN-Cert and POPQORN, respectively for convolutional and recurrent network models. We evaluate the methods on sentiment analysis and news classification models for four datasets (IMDB, SST, AG News and NEWS) and a range of embeddings, and provide an analysis of robustness trends. We also apply our framework to interpretability analysis and compare it with LIME.