Evolutionary Modularity Optimization Clustering of Neuronal Spike Trains
We propose a method for automatic evolutionary clustering of multi neuronal spike trains on the basis of community detection in complex networks. We use a genetic algorithm for optimization to maximize the modularity for community partitioning and then automatically determine the number of clusters hidden in the multi neuronal spike trains. The number of clusters does not need to be specified in advance. Compared with the traditional graph partitioning method, the genetic evolutionary modularity optimization clustering algorithm can obtain the maximum value of modularity and, determine the number of communities. We evaluate the performance of this method on surrogate spike train datasets with ground truth. The results obtained showed improvement. We then apply this proposed method to raw real spike trains. We obtain a larger value for modularity and the results. This finding suggests that the proposed method can be used to detect the hidden firing pattern.
KeywordsSpike trains Modularity Genetic algorithm
This study was supported by the National Natural Science Foundation of China (Project No. 61375122 and Project No. 61572239), China Postdoctoral Science Foundation (Project No. 2014M551324). Scientific Research Foundation for Advanced Talents of Jiangsu University (Project No. 14JDG040).
- 3.Ng, A., Jordan, M., Weiss, Y.: On spectral clustering: analysis and an algorithm. In: Advances in Neural Information Precessing Systems 14 (2001)Google Scholar