Fuzzy-Rough induced spectral ensemble clustering

Ensemble clustering helps achieve fast under abundant computing resources by constructing multiple base clusterings. Compared with the standard single algorithm, ensemble integrates advantages of algorithms and has stronger robustness applicability. Nevertheless, most treat each result equally ignore difference clusters. If a cluster in is reliable/unreliable, it should play critical/uncritical role process. Fuzzy-rough sets offer high degree flexibility enabling vagueness imprecision present real-valued data. In this paper, novel fuzzy-rough induced spectral approach proposed to improve performance clustering. Specifically, significance clusters differentiated, unacceptable reliability formed are based on lower approximation. Based defined reliability, new co-association matrix generated enhance effect diverse Finally, consensus function constructed adjacency matrix, which can lead significantly better results. Experimental results confirm that works effectively outperforms many state-of-the-art clustering, illustrates superiority algorithm.