Cluster Aggregation for Analyzing Event-Related Potentials

Reza Mahini,Tianyi Zhou,Peng Li,Asoke K. Nandi,Huanjie Li,Hong Li,Fengyu Cong

DOI: https://doi.org/10.1007/978-3-319-59081-3_59

2017-01-01

Abstract:Topographic analysis are references independent for Event-Related Potentials (ERPs), and thus render statistically unambiguous results. This drives us to develop an effective clustering approach to finding temporal samples possessing similar topographies for analysing the temporal-spatial ERPs data. The previous study called CARTOOL used single clustering method to cluster ERP data. Indeed, given a clustering method, the quality of clustering varies with data and the number of clusters, motivating us to implement and compare multiple clustering algorithms via using multiple similarity measurements. By finding the minimum distance among the various clustering methods and selecting the most selected clustering algorithms with other methods via voting the proposed method, a most suitable algorithm showing a considerable performance for a given dataset can be found. This cluster aggregation approach assists to use the most suitable founded cluster for each dataset. We demonstrated the effectiveness of the proposed method by using ERP data for cognitive neuroscience research.