Reversible Jump Attack to Textual Classifiers with Modification Reduction

arXiv:2403.14731v1 Announce Type: new
Abstract: Recent studies on adversarial examples expose vulnerabilities of natural language processing (NLP) models. Existing techniques for generating adversarial examples are typically driven by deterministic hierarchical rules that are agnostic to the optimal adversarial examples, a strategy that often results in adversarial samples with a suboptimal balance between magnitudes of changes and attack successes. To this end, in this research we propose two algorithms, Reversible Jump Attack (RJA) and Metropolis-Hasting Modification Reduction (MMR), to generate highly …

adversarial arxiv attack balance cs.cl cs.cr cs.lg examples expose language modification natural natural language natural language processing nlp results rules strategy studies techniques vulnerabilities