A proposal to classify ways of walking patterns using spiking neural networks

Abstract

In this work the Spiking Neural Networks (SNNs) for the classification of ways of walking patterns is presented. The Differential Evolution (DE) Algorithm as an optimization technique was used for weights and delays settings. Two accelerometers, each one with three axes, were used to obtain simultaneous information on both legs. The information formed by nine features has been stored in a database: the first three correspond to the accelerations of x, y and z axis, next three correspond to the velocities which are obtained by doing an integration of the acceleration data for each axis and finally the positions x, y and z are calculated by the integration of velocities respectively.