FastEnsemble: A new scalable ensemble clustering method

Autor:	Tabatabaee, Yasamin, Wedell, Eleanor, Park, Minhyuk, Warnow, Tandy
Rok vydání:	2024
Předmět:	Computer Science - Social and Information Networks
Druh dokumentu:	Working Paper
Popis:	Many community detection algorithms are stochastic in nature, and their output can vary based on different input parameters and random seeds. Consensus clustering methods, such as FastConsensus and ECG, combine clusterings from multiple runs of the same clustering algorithm, in order to improve stability and accuracy. In this study we present a new consensus clustering method, FastEnsemble, and show that it provides advantages over both FastConsensus and ECG. Furthermore, FastEnsemble is designed for use with any clustering method, and we show results using \ourmethod with Leiden optimizing modularity or the Constant Potts model. FastEnsemble is available in Github at https://github.com/ytabatabaee/fast-ensemble Comment: 12 pages, 5 figures, submitted to a conference
Databáze:	arXiv
Externí odkaz:	http://arxiv.org/abs/2409.02077 Zobrazit plný text záznamu View this record from Arxiv