通用指南驱动的图像聚类：一种混合 LLM 智能体框架
Universal Guideline-Driven Image Clustering via a Hybrid LLM Agent
把文本指南作为跨场景聚类的统一接口，并用概念代理生成指南感知的嵌入，这为无监督视觉表征学习提供了一种不依赖任务特定训练就能适配新聚类准则的思路；基于最小生成树的 LLM 遍历则给出了一种只在语义歧义处调用大模型推理的省算力策略。
原文依据 Generative Concept Proxy Modeling, which generates guideline-aware embeddings via concept proxy extraction.
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论由于任务之间存在根本性差异，将不同聚类场景下的图像聚类统一起来仍然具有挑战性。我们提出一种指南驱动的图像聚类智能体，这是首个通过文本指南弥合这些差异的通用框架。为了在不进行任务特定训练的情况下纳入复杂指南，我们提出生成式概念代理建模，通过概念代理提取生成指南感知的嵌入。对于需要自动发现簇的场景，我们引入基于最小生成树的 LLM 遍历，有选择地对复杂语义判断施加 LLM 推理。我们的方法可泛化到从通用到细粒度分类、从全局到局部准则、从均衡到长尾分布等多种聚类场景。我们的框架在多种聚类任务上持续优于专用方法。
英文原文摘要 907 字符
Unifying image clustering across different clustering scenarios remains challenging due to fundamental gaps among tasks. We introduce a Guideline-Driven Image Clustering Agent, the first universal framework that bridges these gaps through textual guidelines. To incorporate complex guidelines without task-specific training, we propose Generative Concept Proxy Modeling, which generates guideline-aware embeddings via concept proxy extraction. For scenarios requiring automatic cluster discovery, we introduce LLM Traversal based on Minimum Spanning Tree that selectively applies LLM reasoning for complex semantic judgments. Our method generalizes across diverse clustering scenarios spanning from general to fine-grained categorization, from global to local criteria, and from balanced to long-tail distributions. Our framework consistently outperforms specialized methods across diverse clustering tasks.