An hybridization of an ant-based clustering algorithm with growing neural gas networks for classification tasks

Conventional ant-based clustering algorithms and growing neural gas networks are combined to produce an unsupervised classification algorithm that exploits the strengths of both techiques. The ant-based clustering algorithm detects existing classes on a training data set, and at the same time, trains several growing neural gas networks. On a second stage, these networks are used to classify previously unseen input vectors into the classes detected by the ant-based algorithm. The proposed algorithm eliminates the need of changing the number of agents and the dimensions of the environment when dealing with large databases.