Exploring Standard Self-Organizing Map Internal Structure in the Learning

In recent years, accumulating evidence suggests that astrocytes type glial cells do much more than supporting neurons. In fact, it is shown that astrocytes communicate, influence and modulate neurons activity, and thus directly affect the information processing in the CNS. In the present article we propose to implement such a modulation in the standard self-organizing map (SOM). We introduce an interactive adaptation/correction behavior by involving SOM´s internal structure in the learning process, allowing the final weight distribution to be invariant to the initial conditions and free of any folding effect due to the well known border effect problem in standard SOM.