On using an ensemble approach of AIS and SVM for text classification

• Type: Article in International Conference Proceedings Book
• Date: 2010

Authors

Mário Antunes (Author) FCUP; Catarina Silva (Author) Other; Bernardete Ribeiro (Author) Other; Manuel Correia (Author) FCUP

Conference proceedings International

Title: Proceedings of 5st Workshop on Applications of Computational Intelligence (WACI)

WACI 10

Coimbra, Portugal, 2 December 2010

Scientific classification

FOS: Natural sciences > Computer and information sciences

CORDIS: Technological sciences > Engineering > Computer engineering

Other information

Abstract (EN): Artificial Immune Systems (AIS) and Support Vector Machines (SVM) are grounded on two radically different conceptual paradigms, each one having intrinsic distinctive features suitable to be successfully applied in dynamic real world applications. One of such applications is the classification of textual documents where each approach individually has proved to obtain promising results. In this paper we aim to present an hybrid system for text classification based on the ensemble of both AIS and SVM approaches. In AIS we explore a binary classification methodology derived from an immunological model which stats that for activation thresholds for T-cells activation is based on the recent history of their iterations with the environment. Regarding the SVM we take advantage of a non-evolutionary implementation that produced remarkable results with text classification. We report some preliminary results on the well-known Reuters-21578 benchmark, showing promising classification performance gains, resulting in a classification that improves upon all baseline contributors of the ensemble committee.

Language: English

Type (Professor's evaluation): Scientific

Notes: ISBN: 978-989-8331-10-6. - IEEE Portuguese Chapter. http://www.isec.pt/eventos/deis/waci10/