Evolutionary clustering detection of similarity in neuronal spike patterns

Abstract

The key to interpreting multi-electrode recorded neuronal spike trains are the firing patterns hidden in a population of neurons. Here, we present a new firing pattern detection method based on community structure partitioning method, in which we apply the genetic evolutionary algorithm to maximize modularity function Q. We propose a new genotype encoding method to represent the functional connections between neurons. Independent of prior ‘knowledge,’ this method automatically finds the number and type of firing patterns in neuronal populations, an advantage over current leading methods.