Consensus Clustering using Partial Evidence Accumulation
; Bulo, S. Bulo
; Rebagliati, N.
Fred, A. L. N.
Figueiredo, M. A. T.
; Pelillo, M.
Consensus Clustering using Partial Evidence Accumulation, Proc Iberian Conf. on Pattern Recognition and Image Analysis, Funchal, Portugal, Vol. ?, pp. ? - ?, June, 2013.
Digital Object Identifier:
The Evidence Accumulation Clustering, EAC, algorithm is a clustering ensemble method which uses co-occurrence statistics to derive a similarity matrix, referred to as co-association matrix. In order to obtain a final consensus clustering the co-association matrix is fed to a pairwise similarity clustering algorithm. The method has thus O(n^2) space complexity, which can constitute a relevant bottleneck to its scalability. In this paper we propose a new formulation which works using a partial set of the co-occurrences, greatly reducing the computational time and space, leading to a scalable algorithm. Experiments on both synthetic and real benchmark data show the effectiveness of the proposed approach.