GLEAN framework uses LLM feedback for generalized category discovery

Researchers have developed GLEAN, a new framework for Generalized Category Discovery (GCD) that utilizes feedback from multiple Large Language Models (LLMs). This approach aims to improve the recognition of both known and novel categories in unlabeled data by enhancing instance-level features, generating category descriptions, and aligning uncertain instances with these descriptions. Experiments show GLEAN outperforms existing state-of-the-art methods across various datasets and settings. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT Introduces a novel method for improving unsupervised learning by leveraging LLM feedback for category discovery.

RANK_REASON This is a research paper detailing a new framework for a machine learning task. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• GLEAN
• LLM
• arXiv

• paper
• other