Study And Application Of Neural Model Based On Olfactory System

A neural network model -- K set model is described in the first part of this thesis. This model is setup according to entire neural architecture of olfactory system based on an amount of electro-psychological experiments. K set model is a high-dimensional chaotic neural network, which can not only simulate EEG waveform observed in experiments, but also perform biological intelligence for pattern classification. It might provide a novel explanation for the signal processing mechanism of the neural system in the aspect of nonlinear dynamics as well.The numeral analysis of the K set model, including KO, KI, KII and KIII model, is presented in this thesis. From the analysis result, it can be seen that KIII network can well simulate the biological olfactory neural system. Different from conventional artificial neural network, the KIII network works in its chaotic trajectory. The learning and pattern recognition process of the network is also relative to its memory basins, which are some kind of chaotic attractors.Based on the understanding of the K set model and previous work, the KIII model is applied on EEG pattern recognition and series of image pattern recognition. The performance of the KIII network is also compared with those results when using conventional artificial neural networks. The pattern recognition results prove that the KIII network is a good pattern classifier.