In evidential clustering, the membership of objects to clusters is considered to be uncertain and is represented by Dempster-Shafer mass functions, forming a credal partition. The EVCLUS algorithm constructs a credal partition in such a way that larger dissimilarities between objects correspond to higher degrees of conflict between the associated mass functions. In this paper, we present several improvements to EVCLUS, making it applicable to very large dissimilarity data. First, the gradient-based optimization procedure in the original EVCLUS algorithm is replaced by a much faster iterative row-wise quadratic programming method. Secondly, we show that EVCLUS can be provided with only a random sample of the dissimilarities, reducing the time and space complexity from quadratic to roughly linear. Finally, we introduce a two-step approach to construct credal partitions assigning masses to selected pairs of clusters, making the algorithm outputs more informative than those of the original EVCLUS, while remaining manageable for large numbers of clusters.

International audienceWe propose a new clustering algorithm based on the evidential K nearest-neighbor (EK-NN) rule. Starting from an initial partition, the algorithm, called EK-NNclus, iteratively reassigns objects to clusters using the EK-NN rule, until a stable partition is obtained. After convergence, the cluster membership of each object is described by a Dempster-Shafer mass function assigning a mass to each cluster and to the whole set of clusters. The mass assigned to the set of clusters can be used to identify outliers. The method can be implemented in a competitive Hopfield neural network, whose energy function is related to the plausibility of the partition. The procedure can thus be seen as searching for the most plausible partition of the data. The EK-NNclus algorithm can be set up to depend on two parameters, the number K of neighbors and a scale parameter, which can be fixed using simple heuristics. The number of clusters does not need to be determined in advance. Numerical experiments with a variety of datasets show that the method generally performs better than density-based and model-based procedures for finding a partition with an unknown number of clusters