arXiv 2602.017032026-02-02cs.LGcs.CL

AGT$^{AO}$：通过对抗门控训练与自适应正交性实现鲁棒且稳定的LLM反向学习

$\textbf{AGT$^{AO}$}$: Robust and Stabilized LLM Unlearning via Adversarial Gating Training with Adaptive Orthogonality

School of Computer Science and Technology, Xi’an Jiaotong University, China（西安交通大学计算机科学与技术学院）；MOE KLINNS Lab, Xi’an Jiaotong University, China（西安交通大学MOE KLINNS实验室）；Shaanxi Province Key Laboratory of Big Data Knowledge Engineering, China（陕西省大数据知识工程重点实验室）

Pengyu Li, Lingling Zhang, Zhitao Gao, Yanrui Wu, Yuxuan Dong, Huan Liu, Bifan Wei, Jun Liu

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

AGT$^{AO}$通过对抗门控训练与自适应正交性，实现LLM反向学习的鲁棒性和实用性平衡。

摘要：

尽管大型语言模型（LLMs）已展现出显著的能力，但它们无意中记忆了敏感数据，这构成了关键的隐私和安全风险。机器反向学习对于减轻这些风险至关重要，但现有的范式面临一个根本性的困境：激进的反向学习往往导致灾难性遗忘，从而降低模型的实用性，而保守的策略则会留下模型易受对抗性恢复攻击的风险。为了解决这一权衡问题，我们提出了AGT$^{AO}$（具有自适应正交性的对抗门控训练），一个统一的框架，旨在协调鲁棒擦除与实用性保持。具体而言，我们的方法引入了自适应正交性（AO），以动态缓解遗忘与保留目标之间的几何梯度冲突，从而最小化意外知识退化。同时，对抗门控训练（AGT）将反向学习建模为潜在空间的min-max游戏，采用基于课程的学习门控机制来模拟并对抗内部恢复尝试。广泛的实验表明，AGT$^{AO}$在反向学习效果（KUR≈0.01）和模型实用性（MMLU 58.30）之间实现了优越的权衡。代码可在https://github.com/TiezMind/AGT-unlearning上获得。

英文摘要：

While Large Language Models (LLMs) have achieved remarkable capabilities, they unintentionally memorize sensitive data, posing critical privacy and security risks. Machine unlearning is pivotal for mitigating these risks, yet existing paradigms face a fundamental dilemma: aggressive unlearning often induces catastrophic forgetting that degrades model utility, whereas conservative strategies risk superficial forgetting, leaving models vulnerable to adversarial recovery. To address this trade-off, we propose $\textbf{AGT$^{AO}$}$ (Adversarial Gating Training with Adaptive Orthogonality), a unified framework designed to reconcile robust erasure with utility preservation. Specifically, our approach introduces $\textbf{Adaptive Orthogonality (AO)}$ to dynamically mitigate geometric gradient conflicts between forgetting and retention objectives, thereby minimizing unintended knowledge degradation. Concurrently, $\textbf{Adversarial Gating Training (AGT)}$ formulates unlearning as a latent-space min-max game, employing a curriculum-based gating mechanism to simulate and counter internal recovery attempts. Extensive experiments demonstrate that $\textbf{AGT$^{AO}$}$ achieves a superior trade-off between unlearning efficacy (KUR $\approx$ 0.01) and model utility (MMLU 58.30). Code is available at https://github.com/TiezMind/AGT-unlearning.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。