Research map / Adversarial robustness and security / Certified and empirical robustness
MultiGuard: Provably Robust Multi-label Classification against Adversarial Examples
Jinyuan Jia, Wenjie Qu, Neil Zhenqiang Gong
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 |
|---|---|---|---|
| include the second terms in the certification inequality (ablation) | MultiGuard · VOC 2007, R=0.5 | certified top-k precision@R: 23.6 -> 31.3 | Improved |
| include the second terms in the certification inequality (ablation) | MultiGuard · VOC 2007, R=0.5 | certified top-k recall@R: 48.8 -> 66.4 | Improved |
| include the second terms in the certification inequality (ablation) | MultiGuard · VOC 2007, R=0.5 | certified top-k f1-score@R: 31.8 -> 42.6 | Improved |
| joint certification over all ground-truth labels instead of per-label (Jia et al.) | MultiGuard · VOC 2007 | certified precision/recall/f1@R | Improved |
| increase k' (labels predicted by base classifier) | MultiGuard · VOC 2007 | certified precision@R | Mixed |
| increase k (labels predicted by smoothed classifier) | MultiGuard · VOC 2007 | certified precision@R | Mixed |
| increase k | MultiGuard · VOC 2007 | certified recall@R | Improved |
| increase noise sigma | MultiGuard · VOC 2007 | certified precision/recall/f1@R | Mixed |
| increase number of noisy samples n | MultiGuard · VOC 2007 | certified precision/recall/f1@R | Improved |
| increase alpha | MultiGuard · VOC 2007 | certified precision/recall/f1@R | No clear effect |
Closest papers by meaning
- PatchDEMUX: A Certifiably Robust Framework for Multi-label Classifiers Against Adversarial Patches (CVPR 2025 · similarity 0.90)
- Characterizing the Evasion Attackability of Multi-label Classifiers (AAAI 2021 · similarity 0.89)
- Certified Robustness to Label-Flipping Attacks via Randomized Smoothing (ICML 2020 · similarity 0.88)
- Domain Knowledge Alleviates Adversarial Attacks in Multi-Label Classifiers (TPAMI 2021 · similarity 0.88)
- Semantic-Aware Multi-Label Adversarial Attacks (CVPR 2024 · similarity 0.87)
- Soften to Defend: Towards Adversarial Robustness via Self-Guided Label Refinement (CVPR 2024 · similarity 0.87)
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.
