Research map / Adversarial robustness and security / Attacks on text and graphs
TIMA: Text-Image Mutual Awareness for Balancing Zero-Shot Adversarial Robustness and Generalization Ability
Fengji Ma, Hei Victor Cheng, Chenxing Li, Li Liu
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 |
|---|---|---|---|
| use TIMA instead of TeCoA | TeCoA · 14-dataset average, PGD eps=8/255 | Zero-shot accuracy: 8.35 -> 21.65 | Improved |
| use TIMA instead of TeCoA | TeCoA · 14-dataset average, PGD eps=1/255 | Zero-shot accuracy: 28.00 -> 32.69 | Improved |
| use TIMA instead of LAAT | LAAT · 14-dataset average, clean | Zero-shot accuracy: 47.78 -> 59.31 | Improved |
| use TIMA instead of PMG | PMG · 14-dataset average, PGD eps=8/255 | Zero-shot accuracy: 5.43 -> 21.65 | Improved |
| train on clean data only with TIMA instead of TeCoA | TeCoA · 14-dataset average, PGD eps=1/255 (no adversarial training) | Zero-shot accuracy: 0.87 -> 11.72 | Improved |
| add ASAM (ablation) | TeCoA · CIFAR10, PGD eps=8/255 | Zero-shot accuracy: 7.06 -> 31.56 | Improved |
| add ASAM (ablation) | TeCoA · CIFAR100, clean | Zero-shot accuracy: 46.00 -> 51.87 | Improved |
| add ASAM (ablation) | TeCoA · ImageNet, clean | Zero-shot accuracy: 43.87 -> 42.75 | Hurt |
| add SC-MHE (ablation) | TeCoA · CIFAR10, PGD eps=8/255 | Zero-shot accuracy: 7.06 -> 23.62 | Improved |
| add SC-MHE (ablation) | TeCoA · CIFAR100, clean | Zero-shot accuracy: 46.00 -> 45.92 | No clear effect |
| add SC-MHE on top of ASAM (full TIMA) (ablation) | TeCoA + ASAM · CIFAR100, clean | Zero-shot accuracy: 51.87 -> 57.90 | Improved |
| add SC-MHE on top of ASAM (full TIMA) (ablation) | TeCoA + ASAM · CIFAR100, PGD eps=8/255 | Zero-shot accuracy: 17.44 -> 14.38 | Hurt |
Closest papers by meaning
- Text-Guided Attention is All You Need for Zero-Shot Robustness in Vision-Language Models (NeurIPS 2024 · similarity 0.89)
- Enhancing CLIP Robustness via Cross-Modality Alignment (NeurIPS 2025 · similarity 0.88)
- Few-Shot Adversarial Prompt Learning on Vision-Language Models (NeurIPS 2024 · similarity 0.87)
- Robustness in Both Domains: CLIP Needs a Robust Text Encoder (NeurIPS 2025 · similarity 0.86)
- Text and Image Are Mutually Beneficial: Enhancing Training-Free Few-Shot Classification with CLIP (AAAI 2025 · similarity 0.86)
- Mining Fine-Grained Image-Text Alignment for Zero-Shot Captioning via Text-Only Training (AAAI 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.
