Research map / Adversarial robustness and security / Black-box and transfer attacks
SSPAttack: A Simple and Sweet Paradigm for Black-Box Hard-Label Textual Adversarial Attack
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use SSPAttack instead of HLGA | BERT hard-label attacks · MR | semantic similarity: 64.6 -> 73.5 | Helped |
| Use SSPAttack instead of HLGA | BERT hard-label attacks · MR | perturbation rate: 13.325 -> 10.615 | Helped |
| Use SSPAttack instead of HLGA | BERT hard-label attacks · AG | semantic similarity: 68.7 -> 73.4 | Helped |
| Use SSPAttack instead of HLGA | BERT hard-label attacks · AG | perturbation rate: 13.325 -> 10.131 | Helped |
| Use SSPAttack instead of HLGA | BERT hard-label attacks · Yahoo | semantic similarity: 71.3 -> 77.3 | Helped |
| Use SSPAttack instead of HLGA | BERT hard-label attacks · Yahoo | perturbation rate: 5.981 -> 4.642 | Helped |
| Use SSPAttack instead of HLGA | BERT hard-label attacks · Yelp | semantic similarity: 78.3 -> 83.7 | Helped |
| Use SSPAttack instead of HLGA | BERT hard-label attacks · Yelp | perturbation rate: 7.039 -> 5.294 | Helped |
| Use SSPAttack instead of HLGA | BERT hard-label attacks · IMDB | semantic similarity: 86.9 -> 90.9 | Helped |
| Use SSPAttack instead of HLGA | BERT hard-label attacks · IMDB | perturbation rate: 3.354 -> 2.452 | Helped |
| Use SSPAttack instead of HLGA | BERT natural-language-inference attacks · SNLI | semantic similarity: 35.4 -> 54.0 | Helped |
| Use SSPAttack instead of TextHoaxer | hard-label attacks on Google Cloud API · Google Cloud API | semantic similarity: 74.8 -> 80.3 | Helped |
Closest papers by meaning
- LimeAttack: Local Explainable Method for Textual Hard-Label Adversarial Attack (AAAI 2024 · similarity 0.91)
- Semantic-Aware Multi-Label Adversarial Attacks (CVPR 2024 · similarity 0.86)
- Adversarial Paraphrasing: A Universal Attack for Humanizing AI-Generated Text (NeurIPS 2025 · similarity 0.86)
- HQA-VLAttack: Towards High Quality Adversarial Attack on Vision-Language Pre-Trained Models (NeurIPS 2025 · similarity 0.84)
- Ask, Attend, Attack: An Effective Decision-Based Black-Box Targeted Attack for Image-to-Text Models (NeurIPS 2024 · similarity 0.84)
- Revisiting Character-level Adversarial Attacks for Language Models (ICML 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.