Visual evoked potential estimation by artificial neural network filter: comparison with the ensemble averaging method

Abstract

The application of an artificial neural network filter (ANNF) to give a non-linear estimation of the visual evoked potential (VEP) is presented. A feed forward ANNF is designed and trained by a training set consisting of a training signal and a target signal. The training signal is the raw VEP from a single trial while the target signal has much higher SNR which is achieved by ensemble averaging of 100 trials. The result shows that 10 ensembles is needed by ANNF to generate a satisfactory result against 60 ensembles required by traditional ensemble averaging. VEP from individual trial could be obtained; thus the study of the variation of signal across trials is possible.