Boosted-Oriented Probabilistic Smoothing-Spline Clustering of Series

TL;DR

This paper introduces a fuzzy clustering method for time series data that eliminates the need for a fuzzifier parameter by combining probabilistic distance clustering with a boosting approach, enhancing clustering performance.

Contribution

The novel approach integrates boosting with probabilistic distance clustering for time series, removing dependence on the fuzzifier parameter and improving unsupervised clustering accuracy.

Findings

Method performs well on various experiments

Eliminates the need for fuzzifier parameter

Improves clustering accuracy for time series

Abstract

Fuzzy clustering methods allow the objects to belong to several clusters simultaneously, with different degrees of membership. However, a factor that influences the performance of fuzzy algorithms is the value of fuzzifier parameter. In this paper, we propose a fuzzy clustering procedure for data (time) series that does not depend on the definition of a fuzzifier parameter. It comes from two approaches, theoretically motivated for unsupervised and supervised classification cases, respectively. The first is the Probabilistic Distance (PD) clustering procedure. The second is the well known Boosting philosophy. Our idea is to adopt a boosting prospective for unsupervised learning problems, in particular we face with non hierarchical clustering problems. The aim is to assign each instance (i.e. a series) of a data set to a cluster. We assume the representative instance of a given cluster (i.e. the cluster center) as a target instance, a loss function as a synthetic index of the global performance and the probability of each instance to belong to a given cluster as the individual contribution of a given instance to the overall solution. The global performance of the proposed method is investigated by various experiments.