A Context-Aware Approach for Textual Adversarial Attack through Probability Difference Guided Beam Search

TL;DR

This paper introduces PDBS, a novel context-aware textual adversarial attack method that uses probability difference guided beam search to improve attack success rates and efficiency over previous models.

Contribution

The paper proposes PDBS, a new attack model that considers all class label probabilities and employs beam search for more effective adversarial attacks.

Findings

PDBS achieves up to +19.5% higher attack success rate.

PDBS outperforms previous models in multiple evaluation metrics.

Extensive experiments and human evaluations confirm PDBS's effectiveness.

Abstract

Textual adversarial attacks expose the vulnerabilities of text classifiers and can be used to improve their robustness. Existing context-aware methods solely consider the gold label probability and use the greedy search when searching an attack path, often limiting the attack efficiency. To tackle these issues, we propose PDBS, a context-aware textual adversarial attack model using Probability Difference guided Beam Search. The probability difference is an overall consideration of all class label probabilities, and PDBS uses it to guide the selection of attack paths. In addition, PDBS uses the beam search to find a successful attack path, thus avoiding suffering from limited search space. Extensive experiments and human evaluation demonstrate that PDBS outperforms previous best models in a series of evaluation metrics, especially bringing up to a +19.5% attack success rate. Ablation studies and qualitative analyses further confirm the efficiency of PDBS.