提出新方法让视觉语言模型彻底遗忘敏感信息，同时避免生成不存在的物体。
Object Hallucination-Free Reinforcement Unlearning for Vision-Language Models
- 在视觉编码器上做强化学习，实现深层语义删除
- 遗忘率超98%，且生成虚假物体极少
- 适合需要隐私保护的VLM应用
视觉语言模型（VLMs）引发对隐私、版权和偏见的担忧，促使机器遗忘技术用于移除敏感知识。然而，现有方法主要微调语言解码器，导致遗忘表层化，无法清除底层视觉表示，常引发物体幻觉。本文提出HFRU，一种基于强化学习的遗忘框架，作用于视觉编码器以实现深层语义删除。该两阶段方法结合对齐破坏与基于GRPO的优化，采用复合奖励，包含抽象奖励以鼓励语义合理替换并抑制幻觉。在物体识别和人脸身份任务上的实验表明，HFRU实现超过98%的遗忘率与保留性能，同时引入可忽略的物体幻觉，显著优于先前方法。代码与实现细节见https://github.com/XMUDeepLIT/HFRU。
原文摘要 · Abstract (English)
Vision-language models (VLMs) raise growing concerns about privacy, copyright, and bias, motivating machine unlearning to remove sensitive knowledge. However, existing methods primarily fine-tune the language decoder, leading to superficial forgetting that fails to erase underlying visual representations and often introduces object hallucination. We propose HFRU, a reinforcement unlearning framework that operates on the vision encoder for deep semantic removal. Our two-stage approach combines alignment disruption with GRPO-based optimization using a composite reward, including an abstraction reward that encourages semantically valid substitutions and mitigates hallucinations. Experiments on object recognition and face identity tasks show that HFRU achieves over 98% forgetting and retention performance, while introducing negligible object hallucination, significantly outperforming prior methods.Our code and implementation details are available at https://github.com/XMUDeepLIT/HFRU.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。