A Clustering Algorithm Based on the Ants Self-Assembly Behavior

We have presented in this paper an ants based clustering algorithm which is inspired from the self-assembling behavior observed in real ants. These ants progressively become connected to an initial point called the support and then successively to other connected ants. The artificial ants that we have defined similarly build a tree where each ant represents a node/data. Ants use the similarities between the data in order to decide where to connect. We have tested our method on numerical databases (either artificial, real, and from the CE.R.I.E.S.). We show that AntTree improves the clustering process compared to the Kmeans algorithm and to AntClass, a previous approach for data clustering with artificial ants.