Using a Kalman filter that contains a forward-predictive model of a relevant system, to predict the states of that system by means of an analysis-by-synthesis implementation in order to evade significant time delays incurred by feedback mechanisms was previously applied to the coordinated movement of limbs by means of the cerebellum. In this dissertation, the same concept was applied to the auditory system in order to investigate if such a concept is a universal neurophysiological method for correctly estimating a state in a quick and reliable way. To test this assumption an auditory system model and Kalman estimator were designed, where the Kalman filter contained a stochastically equivalent forward-predictive model of the complete auditory system model. The Kalman filter was used to estimate the power found in a particular band of the frequency spectrum and its performance in the mean-squared error sense was compared to that of a simple postsynaptic current decoding filter under various types of neural channel noise. It was shown that the Kalman filter, containing a biologically plausible internal model could estimate the power better than a postsynaptic current decoding filter, proposed in the literature. When the just-noticeable difference in intensity discrimination, as reported in the literature, was compared to model-predictions, it was shown that a smaller mean-squared error results in the case of the designed auditory system model and Kalman estimator. This suggests that the application of the Kalman filter concept is important as it provides a bridge between measured data and the auditory system model. It was concluded that a Kalman filter model containing a biologically plausible internal model can explain some characteristics of the signal processing of the auditory system. The research suggests that the principle of an estimator that contains an internal model could be a universal neurophysiological method for the correct estimation of a desired state. / Dissertation (MEng (Electronic Engineering))--University of Pretoria, 2006. / Electrical, Electronic and Computer Engineering / unrestricted
| Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:up/oai:repository.up.ac.za:2263/25392 |
| Date | 10 June 2005 |
| Creators | Hauger, Martin M |
| Contributors | Prof J Hanekom, samadhound@hotmail.com |
| Source Sets | South African National ETD Portal |
| Detected Language | English |
| Type | Dissertation |
| Rights | © 2005, University of Pretoria. All rights reserved. The copyright in this work vests in the University of Pretoria. No part of this work may be reproduced or transmitted in any form or by any means, without the prior written permission of the University of Pretoria. |
Page generated in 0.0028 seconds