Volterra Series Analysis of a Neural Network for Velocity Estimation

2. THE MOTION DETECTION MODEL In this paper we propose and analyze a motion detection Neural networks that are based on modulation of con-circuit which is based on nerve membrane conduction. It ductance of nerve membranes by other cells of the network consists of two unidirectional neural networks connected have been studied extensively and found to have widely in an opposing fashion. Volterra input-output models are varied and complex computational properties such as then derived for the network so that the velocity estima-contrast enhancement, dynamic range compression, and tion can then be cast as a parameter estimation problem. adaptation to mean input levels (for a review see [11, 12]). The technique is demonstrated through simulation. The elementary shunting motion detector (ESMD) is an intensity, or luminescence, based scheme of short range