Statistical Modeling of an Electronic Olfactory

Abstract

This thesis attempts to provide an understanding of the natural design principles that underlie the observed learning/clustering performance of the modified GLA olfactory neural network which retains the essential clustering properties of the olfactory bulb and paleocortex in pattern recognition. A statistical model is developed to model the proposed hardware implementation of the modified GLA model. This statistical modelling of the modified GLA model will assist in the understanding and optimizing the design and architecture dimensionality and also in intepreting the test results.