Association analysis of biosignals using self organizing maps

Abstract

This work assesses the ability of Self Organizing Maps (SOMs) to find nonlinear association and/or connectivity among biosignals. The proposed method can find numerous applications where nonlinear biosignals are measured in spatiotemporal manner. Experiments are performed on tens of thousands of biosignals that are obtained from real biosignals by implementing a nonlinear transform, delays, additive and multiplicative random noise. Results showed that resolving association among biosignals under strong nonlinear transformation, noise, and delay is effective using SOMs.