Simulation of neural network in autism

Abstract

This report presents a project in artificial neural network. It aims to create a neurocomputational model for autism. The specific neural network implemented will be Self Organization Maps (SOM). Such topic has been previously investigated by others. The projects aim to verify conclusion drawn from past developments and characteristic of autism through simulation of neural network in autism. Familiarity inclination and attention shift impairment, features identified in autism will be simulated and program outputs will be discussed in details. Result corresponding to cortical map viewed in normal human being will be explored and compared with autism traits simulated. This report will also be covering the impact of contrasting values used for learning rate, training length, data input format and neighbouring function on cortical maps generated.