Research map / Adversarial robustness and security / Attacks on text and graphs
Multi-Expert Adversarial Attack Detection in Person Re-Identification Using Context Inconsistency
Xueping Wang, Shasha Li, Min Liu, Yaonan Wang, Amit K. Roy-Chowdhury
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| add more expert models (one to four) (ablation) | MEAAD · Market1501 | detection F1: 95.5 -> 98.6 | Improved |
| add the attack target model to the other three experts (ablation) | MEAAD · Market1501 | detection F1: 94.2 -> 98.6 | Improved |
| increase support set size from K=1 to K=15 (ablation) | MEAAD · Market1501 | detection accuracy: 92.3 -> 98.5 | Improved |
| increase support set size beyond K=15 (ablation) | MEAAD · Market1501 | detection accuracy: 98.5 -> 98.5 | No clear effect |
| add cross-expert affinity to query-support and support-support (ablation) | MEAAD · Market1501 | detection F1: 97.0 -> 98.6 | Improved |
| add query-support affinity to cross-expert and support-support (ablation) | MEAAD · Market1501 | detection F1: 95.7 -> 98.6 | Improved |
| add support-support affinity to cross-expert and query-support (ablation) | MEAAD · Market1501 | detection F1: 97.1 -> 98.6 | Improved |
| replace threshold voting on common retrievals with a trained MLP detector (ablation) | MEAAD · Market1501 (Deep Mis-Ranking) | detection F1: 91.0 -> 98.6 | Improved |
| test detector on unseen Deepfool attack instead of the training attack CW (ablation) | MEAAD · Market1501 | detection F1: 96.2 -> 95.9 | No clear effect |
| train detector on PGD and test on unseen CW attack (ablation) | MEAAD · Market1501 | detection F1: 97.4 -> 93.8 | Hurt |
| adaptive CW attack with context-affinity loss term (target model white-box) | MEAAD · Market1501 | detection accuracy: -1.3% | Hurt |
| adaptive CW attack with context-affinity loss term (all models white-box) | MEAAD · Market1501 | detection accuracy: -3.5% | Hurt |
Closest papers by meaning
- Transferable, Controllable, and Inconspicuous Adversarial Attacks on Person Re-identification With Deep Mis-Ranking (CVPR 2020 · similarity 0.91)
- Debiased Dual-Invariant Defense for Adversarially Robust Person Re-Identification (AAAI 2026 · similarity 0.89)
- R$^2$TUA: Reconstruction-residual Based Targeted and Untargeted Attack Against Text-Image Person Re-Identification (CVPR 2026 · similarity 0.89)
- Towards Robust Person Re-Identification by Defending Against Universal Attackers (TPAMI 2022 · similarity 0.88)
- Adversarially Occluded Samples for Person Re-Identification (CVPR 2018 · similarity 0.87)
- Feature-Level Adversarial Attacks and Ranking Disruption for Visible-Infrared Person Re-identification (NeurIPS 2024 · similarity 0.86)
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.
