AI Paper Atlas

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

DART: Distribution-Aware Adaptive Relational Transfer for Adversarial Attacks against Closed-Source MLLMs

Kaidi Hu, Guancheng Wan, Xiao Luo, Ruigang Yang

ICML 2026 · Adversarial robustness and security › Attacks on text and graphs · approach: Attacks and defenses in sequential settings

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 Energy Distance distribution alignment with M=5 augmentations (vs point-wise M=1) (ablation)DART · GPT-4o ablationASR: 83.0 -> 89.5Improved
add intra-distribution terms (Energy Distance vs AvgCosine/EoT) (ablation)DART · GPT-4o ablationASR: 82.5 -> 89.5Improved
add data augmentation with simple gradient smoothing only (AvgCosine, M=5) vs point-wise (ablation)point-wise attack · GPT-4o ablationASR: 83.0 -> 82.5No clear effect
use Energy Distance instead of MMD (ablation)DART · GPT-4oASRImproved
increase GNN router ensemble size from K=1 to K=2 (ablation)DART · GPT-4oASRImproved
increase ensemble size from K=2 to K=3 (ablation)DART · GPT-4oASRHurt
increase augmented samples M from 2 to 5 (ablation)DART · GPT-4oASR: +3.0% from M=1 to M=2; saturates beyond M=5Improved
use heterogeneous global/local weighting eta=[1.8, 2.1, 4.8] instead of uniform (ablation)DART · GPT-4oASRImproved
replace FOA-Attack with DART on open-source Qwen2.5-VL-7BDART · Open-source MLLMsASR: 80.0 -> 86.5Improved
replace FOA-Attack with DART on Claude-3.7DART · Closed-source MLLMsASR: 23.5 -> 32.0Improved
apply Gaussian blur defense (r=1.0) to adversarial imagesDART vs FOA-Attack · GPT-4oASR: 89.5 -> 65.0 (FOA 85.5 -> 53.0)Hurt
reduce M from 5 to 2 for efficiency (ablation)DART · GPT-4oASR: 89.5 -> 86.0; time 175 s -> 86 sMixed

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