Research Seminar

A novel approach for fitting the INDCLUS model


Dirk Depril


KU Leuven

Abstract: In this talk we will consider N x N x K data, the K slices
consisting of N x N similarity matrices. A well-known clustering method
to analyze such data is the INDCLUS model (a three way version of
ADCLUS), which clusters the N objects into overlapping clusters
according to their similarities over the K slices. During the past three
decades, several authors have proposed an algorithm to fit the INDCLUS
model to a given data set. However, all these algorithms are either
computationally intensive or suffer from a local optima problem. In this
talk we will propose a novel approach for fitting the model. We will
evaluate it in a simulation study and discuss the results
Date: Tue Dec 4, 12:15 pm - 1:15 pm
Place: room 00.60 (Department of Psychology, Tiensestraat 102, 3000 Leuven)