PGPO: Prototype-Guided Progressive Obfuscation for Privacy-Preserving LLM-Enhanced Recommendation

媒体报道 Hugging Face Papers 原文语言:中文

原文摘录

论文摘要(作者报告,未经本站复测):LLM-enhanced recommender systems encode item side information into semantic embeddings, but these embeddings may leak sensitive textual attributes through inversion attacks. Existing input-side defenses rely on trusted collection or local token perturbation, often disrupting the inter-item semantic topology needed for recommendation.

以下是来源提供的简短摘录,完整内容请查看原文。

参考来源

← 返回每日AI报告