New framework evaluates clustering by homogeneity-parsimony trade-off

A new paper introduces a framework for evaluating clusterings against known classes by focusing on the trade-off between homogeneity and parsimony. The proposed scores, derived from the information bottleneck principle, quantify how informative a clustering is about class labels without excessive fragmentation. This approach offers a unified view of common evaluation criteria and can be applied to tasks like feature selection and algorithm comparison. AI

IMPACT Provides a novel method for evaluating clustering performance, potentially improving feature selection and algorithm development in machine learning.

RANK_REASON The cluster contains a research paper detailing a new framework for evaluating clustering algorithms. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• alphaXiv
• Andreas Tiffeau-Mayer
• arXiv
• CatalyzeX Code Finder for Papers
• CORE Recommender
• DagsHub
• Gotit.pub
• Homogeneity-Parsimony Trade-off
• Hugging Face
• IArxiv Recommender
• receiver operating characteristic
• ScienceCast
• Shannon entropies

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →