A novel and simple strategy for evolving prototype based clustering. (October 2018)

• Record Type: Journal Article
• Authors: Márquez, David G.; Otero, Abraham; Félix, Paulo; García, Constantino A.
• Is Part Of: Pattern recognition. Volume 82(2018:Oct.)
• Page Start: 16
• Page End: 30
• Publication Date: 2018-10
• Subjects: Evolving clustering -- Data stream -- Concept drift -- Gaussian mixture models -- K-means -- Cluster evolution; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2018.04.020
• Languages: English
• ISSNs: 0031-3203

Abstract

Highlights: A novel strategy to evolve clusters by gradually forgetting old samples is proposed. It is based on a dynamic weighting schema with an adjustable "memory" parameter. It was used to develop evolving versions of K-means and Mixture of Models. The algorithms are specially geared towards data drift scenarios. They were tested over real data and a synthetic data drift oriented database. Abstract: In this paper, we present a novel strategy for evolving prototype based clusters that uses a weighting scheme to "progressively forget" old samples. The rate of forgetfulness can be controlled by a single intuitive memory parameter. This weighting scheme can be used to create efficient dynamic summaries, such as mean or covariance, of data streams. Using this weighting scheme we have developed evolving versions of the K-means and Gaussian Mixture models algorithms. They can analyze the incoming data in an online manner and they are specially geared towards dealing with concept drift originated by changes in the underlying data distribution. The algorithms were validated over a simulated database where a wide variety of concept drift situations occur and over real data related to property sales, showing their capability to follow changes in data.