Breaking Cross-modal Alignment in Embodied Intelligence: A Multimodal Adversarial Attack Framework for Vision-Language-Action Models
Open Access
- 12 April 2026
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 2824-2834
Abstract
No abstract availableFunding Information
- National Natural Science Foundation of China (62472302)
- Key Research and Development Program of Shanxi Province (202402150101007)
- Open Project of the State Key Laboratory of Internet Architecture (HLW2025MS31)
This publication has 14 references indexed in Scilit:
- Accuracy is not enough: a heterogeneous ensemble model versus FGSM attackComplex & Intelligent Systems, 2024
- Vision-Language Models for Vision Tasks: A SurveyIEEE Transactions on Pattern Analysis and Machine Intelligence, 2024
- Performance Study of CBAM Attention Mechanism in Convolutional Neural Networks at Different Depths2023
- Invertible Residual Blocks in Deep Learning NetworksIEEE Transactions on Neural Networks and Learning Systems, 2023
- Mean Squared Error, DeconstructedJournal of Advances in Modeling Earth Systems, 2021
- Oscar: Object-Semantics Aligned Pre-training for Vision-Language Tasks2020
- 3D Instance Embedding Learning With a Structure-Aware Loss Function for Point Cloud SegmentationIEEE Robotics and Automation Letters, 2020
- Noise is Inside Me! Generating Adversarial Perturbations with Noise Derived from Natural Filters2020
- Local Gradients Smoothing: Defense Against Localized Adversarial Attacks2019
- U-Net: Convolutional Networks for Biomedical Image Segmentation2015