AI Paper Atlas

Research map / Adversarial robustness and security / Attacks on text and graphs

Stabilizing Cross-Modal Bidirectional Attribution: Few-Shot Adversarial Prompt Tuning for Robust Vision-Language Models

Jun Feng, Shuhong Wu, Hong Sun, Pengfei Zhang, Bocheng Ren, Shunli Zhang

AAAI 2026 · Adversarial robustness and security › Attacks on text and graphs · approach: Vision-language model robustness · 0 citations (OpenAlex)

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 changedWhereMeasuredEffect
add natural cross-entropy loss to adversarial loss (ablation)CBA-FAPT · Base-to-new (16-shot, avg 11 datasets)New natural accuracy (%): 46.18 -> 51.74Improved
add natural cross-entropy loss to adversarial loss (ablation)CBA-FAPT · Base-to-new (16-shot, avg 11 datasets)Base robust accuracy (%): 30.69 -> 35.72Improved
add visual attribution alignment loss (ablation)CBA-FAPT · Base-to-new (16-shot, avg 11 datasets)New natural accuracy (%): 51.74 -> 53.42Improved
add visual attribution alignment loss (ablation)CBA-FAPT · Base-to-new (16-shot, avg 11 datasets)New robust accuracy (%): 23.38 -> 24.17Improved
add textual attribution alignment (full cross-modal alignment) (ablation)CBA-FAPT · Base-to-new (16-shot, avg 11 datasets)New robust accuracy (%): 24.17 -> 25.44Improved
add textual attribution alignment (full cross-modal alignment) (ablation)CBA-FAPT · Base-to-new (16-shot, avg 11 datasets)Base robust accuracy (%): 36.45 -> 38.22Improved
full CBA-FAPT vs FAP on new classesCBA-FAPT · Base-to-new (16-shot, avg 11 datasets)New natural accuracy (%): 49.58 -> 55.74Improved
full CBA-FAPT vs FAP on new classesCBA-FAPT · Base-to-new (16-shot, avg 11 datasets)New robust accuracy (%): 21.86 -> 25.44Improved
full CBA-FAPT vs FAP on base classesCBA-FAPT · Base-to-new (16-shot, avg 11 datasets)Base natural accuracy (%): 70.52 -> 70.74No clear effect
CBA-FAPT vs FAP, cross-dataset natural accuracyCBA-FAPT · Cross-dataset (ImageNet 16-shot, avg 11 datasets)Natural accuracy (%): 47.76 -> 49.67Improved
CBA-FAPT vs FAP, cross-dataset robust accuracyCBA-FAPT · Cross-dataset (ImageNet 16-shot, avg 11 datasets)Robust accuracy (%): 24.23 -> 24.69Improved
CBA-FAPT vs FAP on Caltech101 robust accuracyCBA-FAPT · Cross-dataset (Caltech101)Robust accuracy (%): 65.43 -> 63.70Hurt

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy