ECAI 2004 Conference Paper

[PDF] [full paper] [prev] [tofc] [next]

PoBOC : an overlapping clustering algorithm. Application to classification rules learning

Guillaume Cleuziou, Lionel Martin, Christel Vrain

This paper presents the clustering algorithm PoBOC (Pole-Based Overlapping Clustering). It has two main characteristics: the number of final clusters is unknown a priori and PoBoc allows an object to belong to one or several clusters. Starting from a similarity matrix over a set of objects, it first builds a small and homogeneous sets of objects (the poles), and then it assigns the objects to the poles. PoBOC differs from traditional clustering methods which aim at maximizing the intra-cluster similarity and minimizing the inter-cluster similarity. One of its main applications concern the pre-processing task which allows the organization of data in a knowledge discovery perspective. The clustering method is evaluated on the Rule-Based Learning (RBL) task. Classification rules are generated by organizing the instances of a class so that each cluster is covered with a single rule. PoBOC is compared to different clustering methods on traditional datasets from the UCI repository. We show that our approach is appropriated to the RBL task, with respect to other clustering technics. Moreover, the global classification method leads, on some datasets, to better results than famous classifiers such as decision-tree or nearest neighbour ones.

Keywords: Machine Learning, Clustering, Rule-Based Learning, Data Mining

Citation: Guillaume Cleuziou, Lionel Martin, Christel Vrain: PoBOC : an overlapping clustering algorithm. Application to classification rules learning. In R.López de Mántaras and L.Saitta (eds.): ECAI2004, Proceedings of the 16th European Conference on Artificial Intelligence, IOS Press, Amsterdam, 2004, pp.440-444.

[prev] [tofc] [next]

ECAI-2004 is organised by the European Coordinating Committee for Artificial Intelligence (ECCAI) and hosted by the Universitat Politècnica de València on behalf of Asociación Española de Inteligencia Artificial (AEPIA) and Associació Catalana d'Intel-ligència Artificial (ACIA).