The Time Adaptive Self-Organizing Map is a Neural Network Based on Artificial Immune System

In this paper, the similarities between the mechanisms used in the TASOM (time adaptive self-organizing map) neural network and AIS (artificial immune systems) are analyzed. To demonstrate the similarities, AIS mechanisms are incorporated into the TASOM network such as the weight updating is replaced by a mutation mechanism. Learning rate and neighborhood sizes are also replaced by the clonal selection process used in AIS. This new network is called TAISOM. Experimental results with TAISOM are implemented for uniform and Gaussian distributions for one and two-dimensional lattices of neurons. These experiments show that TAISOM learns its environment as expected so that neurons fill the environments quite well and the neurons also preserve the topological ordering.