A Mixture Model for Spike Train Ensemble Analysis Using Spectral Clustering

Identifying clusters of neurons with correlated spiking activity in large-size neuronal ensembles recorded with high-density multielectrode array is an emerging problem in computational neuroscience. We propose a nonparametric approach that represents multiple neural spike trains by a mixed point process model. A spectral clustering algorithm is applied to identify the clusters of neurons through their correlated firing activities. The advantage of the proposed technique is its ability to efficiently identify large populations of neurons with correlated spiking activity independent of the temporal scale. We report the clustering performance of the algorithm applied to a complex synthesized data set and compare it to multiple clustering techniques