Output details
11 - Computer Science and Informatics
University of Glasgow
Applying the possibilistic C-means algorithm in kernel-induced spaces
<24>This paper proposes a novel algorithm based on clustering and kernel methods, to cluster data sets without imposing assumptions on the shape and the number of the clusters.
The paper analyses a special case that shares similarities with other popular algorithms for novelty detection, clustering, and density estimation, such as One-Class Support Vector Machines and Kernel Density Estimation. Such analysis suggests a number of advantages of the proposed approach in correctly identifying outliers compared to One-Class Support Vector Machines.
This paper was published in one of the top journal in the field of fuzzy systems and artificial intelligence.