Action potential classification based on LVQ neural network

Abstract

Neural action potential classification is the prerequisite condition of further neural information process, but there are difficulties in accurate action potential classification due to the existence of great amount of noise. In this paper, we propose a method of action potential classification based on Learning Vector Quantization (LVQ) network. In the experimental stage, the performance of the presented system was tested at various signal-to-noise ratio levels based on synthetic data. The results show that the proposed action potential classification method is effective. The proposed method supplies a new thought for action potential classification problem. © 2010 Springer-Verlag Berlin Heidelberg.