MLKD logo   Machine Learning &
Knowledge Discovery Group

Publication Details

  Author(s): I. Katakis, G. Tsoumakas, I. Vlahavas.

Title: “An Ensemble of Classifiers for coping with Recurring Contexts in Data Streams”.

Availability: Click here to download the PDF (Acrobat Reader) file.

Keywords: text classification, ensemble, data stream, concept drift, classification.

Appeared in: 18th European Conference on Artificial Intelligence, IOS Press, pp. 763-764, Patras, Greece, 2008.

Abstract: This paper proposes a general framework for classifying data streams by exploiting incremental clustering in order to dynamically build and update an ensemble of incremental classifiers. To achieve this, a transformation function that maps batches of examples into a new conceptual feature space is proposed. An incremental clustering algorithm is then applied in order to group different concepts and identify reoccurring themes. The ensemble is produced by creating and training an incremental classifier for every concept discovered in the data stream. An experimental study is performed using three new real-world datasets from the text domain, a basic implementation of the proposed framework and three baseline methods for dealing with drifting concepts. Results are promising and encourage further investigation.

Relevant Links: