Self-organising sensory maps in odour classification mimicking

Abstract

A system for artificial olfaction is introduced, which is composed of a sensor array for gas sensing and a self-organising artificial neural network. A detailed reformulation of the most effective Self-Organising sensory Map (SOM)-based algorithms for odour classification and other applications is provided. An opto-electronic micromachined implementation of the neural network is introduced, which employs a novel hybrid mechanism for activating neural groups, avoiding fabricated cloning templates hardware.