![[2602.22227] To Deceive is to Teach? Forging Perceptual Robustness via Adversarial Reinforcement Learning](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.22227] To Deceive is to Teach? Forging Perceptual Robustness via Adversarial Reinforcement Learning
Summary
The paper introduces AOT-SFT, an adversarial dataset aimed at enhancing the robustness of Multimodal Large Language Models (MLLMs) against perceptual challenges through a self-play framework.
Why It Matters
As MLLMs face limitations due to finite training datasets, this research provides a novel approach to improve their robustness. By leveraging adversarial training, it addresses critical issues like hallucinations and perceptual fragility, which are vital for the deployment of reliable AI systems.
Key Takeaways
- AOT-SFT is a large-scale adversarial dataset designed to improve MLLM robustness.
- The proposed AOT method facilitates a self-play framework for continuous learning.
- Adversarial training reduces hallucinations in MLLMs, enhancing reliability.
- The approach fosters a dynamic curriculum of image manipulations for better adaptability.
- This research paves the way for scalable training paradigms in AI.
Computer Science > Machine Learning arXiv:2602.22227 (cs) [Submitted on 24 Jan 2026] Title:To Deceive is to Teach? Forging Perceptual Robustness via Adversarial Reinforcement Learning Authors:Yicheng Bao, Xuhong Wang, Xin Tan View a PDF of the paper titled To Deceive is to Teach? Forging Perceptual Robustness via Adversarial Reinforcement Learning, by Yicheng Bao and 2 other authors View PDF HTML (experimental) Abstract:Despite their impressive capabilities, Multimodal Large Language Models (MLLMs) exhibit perceptual fragility when confronted with visually complex scenes. This weakness stems from a reliance on finite training datasets, which are prohibitively expensive to scale and impose a ceiling on model robustness. We introduce \textbf{AOT-SFT}, a large-scale adversarial dataset for bootstrapping MLLM robustness. Building on this, we propose \textbf{AOT (Adversarial Opponent Training)}, a self-play framework that forges MLLM robustness by creating its own training data. Our method orchestrates a co-evolution between an image-editing Attacker and a Defender MLLM, where the Attacker generates a diverse and dynamic curriculum of image manipulations, forcing the Defender to adapt and improve. Extensive experiments demonstrate that AOT enhances the Defender's perceptual robustness and reduces hallucinations, establishing a scalable paradigm for training more reliable MLLMs. Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI) Cite as: arXiv:2602.22227 [cs.LG]...
