ARMOR: Agentic Reasoning for Methods Orchestration and Reparameterization for Robust Adversarial Attacks
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866911399383400448 |
|---|---|
| author | Rong, Gabriel Lee Jun Korgialas, Christos Ho, Dion Jia Xu Ng, Pai Chet Miao, Xiaoxiao Plataniotis, Konstantinos N. |
| author_facet | Rong, Gabriel Lee Jun Korgialas, Christos Ho, Dion Jia Xu Ng, Pai Chet Miao, Xiaoxiao Plataniotis, Konstantinos N. |
| contents | Existing automated attack suites operate as static ensembles with fixed sequences, lacking strategic adaptation and semantic awareness. This paper introduces the Agentic Reasoning for Methods Orchestration and Reparameterization (ARMOR) framework to address these limitations. ARMOR orchestrates three canonical adversarial primitives, Carlini-Wagner (CW), Jacobian-based Saliency Map Attack (JSMA), and Spatially Transformed Attacks (STA) via Vision Language Models (VLM)-guided agents that collaboratively generate and synthesize perturbations through a shared ``Mixing Desk". Large Language Models (LLMs) adaptively tune and reparameterize parallel attack agents in a real-time, closed-loop system that exploits image-specific semantic vulnerabilities. On standard benchmarks, ARMOR achieves improved cross-architecture transfer and reliably fools both settings, delivering a blended output for blind targets and selecting the best attack or blended attacks for white-box targets using a confidence-and-SSIM score. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_18386 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | ARMOR: Agentic Reasoning for Methods Orchestration and Reparameterization for Robust Adversarial Attacks Rong, Gabriel Lee Jun Korgialas, Christos Ho, Dion Jia Xu Ng, Pai Chet Miao, Xiaoxiao Plataniotis, Konstantinos N. Computer Vision and Pattern Recognition Existing automated attack suites operate as static ensembles with fixed sequences, lacking strategic adaptation and semantic awareness. This paper introduces the Agentic Reasoning for Methods Orchestration and Reparameterization (ARMOR) framework to address these limitations. ARMOR orchestrates three canonical adversarial primitives, Carlini-Wagner (CW), Jacobian-based Saliency Map Attack (JSMA), and Spatially Transformed Attacks (STA) via Vision Language Models (VLM)-guided agents that collaboratively generate and synthesize perturbations through a shared ``Mixing Desk". Large Language Models (LLMs) adaptively tune and reparameterize parallel attack agents in a real-time, closed-loop system that exploits image-specific semantic vulnerabilities. On standard benchmarks, ARMOR achieves improved cross-architecture transfer and reliably fools both settings, delivering a blended output for blind targets and selecting the best attack or blended attacks for white-box targets using a confidence-and-SSIM score. |
| title | ARMOR: Agentic Reasoning for Methods Orchestration and Reparameterization for Robust Adversarial Attacks |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2601.18386 |