LBA: Textual Hard-Label Adversarial Attack under Low Query Budgets
Quick Answer
The LBA method introduces a sampling-based approach for generating high-quality adversarial texts under low query budgets, outperforming state-of-the-art techniques across six language models and four datasets.
Quick Take
The LBA method introduces a sampling-based approach for generating high-quality adversarial texts under low query budgets, outperforming state-of-the-art techniques across six language models and four datasets. By integrating prior and posterior knowledge, LBA enhances sampling efficiency, resulting in more semantically coherent adversarial examples.
Key Points
- LBA significantly outperforms existing methods in generating adversarial texts.
- The method integrates prior and posterior knowledge for improved sampling.
- Extensive experiments were conducted on six language models across four datasets.
- LBA generates adversarial examples that are more semantically preserved and comprehensible.
- The approach reduces excessive query costs associated with traditional greedy algorithms.
Paper Resources
📖 Reader Mode
~2 min readAbstract:Generating high-quality adversarial texts with low query budgets remains a challenging problem in the hard-label scenario. Most existing approaches rely on greedy algorithms, where one position in the text is selected for substitution, followed by the substitutions of other positions. This local search approach may fail to discover high-quality adversarial examples and often leads to excessive query costs. Ideally, an optimal adversarial sample would consider all possible position combinations in the text, but exhaustive search is computationally impractical. To address this challenge, we propose a sampling-based method called LBA, which constructs an approximate distribution of high-quality adversarial examples by integrating both prior and posterior knowledge, and utilizes this distribution for sampling. As sampling progresses, posterior knowledge updates the approximate distribution, which in turn guides more effective sampling. Extensive experiments on six language models, ranging from small-scale to large-scale architectures across four datasets, demonstrate that LBA significantly outperforms state-of-the-art baselines on all evaluation metrics. Additionally, LLM-based assessment indicates that LBA generates more semantically preserved and comprehensible adversarial texts.
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2607.14101 [cs.CL] |
| (or arXiv:2607.14101v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2607.14101 arXiv-issued DOI via DataCite |
Submission history
From: Shixin Guo [view email]
[v1]
Tue, 5 May 2026 13:40:10 UTC (2,761 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CL
See more →Letting the Data Speak: Extracting Keywords from Crowdsourced Collections with AI
The study evaluates three NLP approaches—Named Entity Recognition, Keyword Extraction, and Topic Modelling—using the Their Finest Hour Online Archive to automate keyword extraction from crowdsourced WWII collections. Findings suggest that while NLP methods show promise, no single approach is sufficient, and ethical considerations in automated keyword extraction are crucial for responsible stewardship.