Reproduce “k-HyperEdge Medoids for Clustering Ensemble” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Feijiang Li, Jieting Wang, Liuya zhang, Yuhua Qian, Shuai jin, Tao Yan, Liang Du
The Clustering Ensemble based on k-HyperEdge Medoids (CEHM) method, developed by researchers at Shanxi University, reformulates clustering ensemble as a hypergraph problem, combining cluster-level efficiency with sample-level refinement. It demonstrated superior or comparable performance across 20 benchmark datasets in NMI and ARI, achieving high computational efficiency.
Key insights
- Formulating clustering ensemble as a k-HyperEdge Medoids discovery problem provides a flexible and powerful hypergraph-based representation for complex relationships among clusters and samples.
- A hybrid strategy that first efficiently selects cluster-level representatives and then refines them through sample-level interactions can effectively mitigate the limitations of both traditional clustering-view and sample-view methods.
- The defined 'belonging degree' metric offers a statistically sound and robust measure for guiding sample assignments and refining hyperedges during the iterative consensus formation process.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.