Diffusion and Growing Self-Organizing Map: A Nitric Oxide Based Neural Model

This paper presents a new diffusion and growing neural network model called DGSOM for self-organization, where more generalized ideas of short-range competition and long-range cooperation are adopted while introducing the diffusion mechanism of intrinsic NO as its most remarkable characteristic. The new DGSOM model can compartmentalize input space rationally and efficiently, and generated topological connections among neurons can reflect the dimensionality and structure of input signals. Experiments and simulations indicated that the embedding of NO diffusion mechanism improve the performance of self-organization remarkably, especially the flexibility and rapidity of response for dynamic distribution.