AB Distance Based Histogram Clustering for Mining Multi-Channel EEG Data Using Wavesim Transform

Temporal data mining is concerned with the analysis of temporal data and finding temporal patterns, regularities, trends, clusters in sets of temporal data. In this paper we extract histogram features from the coefficients obtained by applying WaveSim transform on multi-channel signals. WaveSim transform is a reverse approach for generating wavelet transform like coefficients by using a conventional similarity measure between the function fit and the wavelet. We propose a method for histogram clustering based on AB distance measure which is based on the `area' and `behavior difference' components between the regression lines obtained from the histograms. The distance measure is used for k-means histogram clustering. WaveSim transform provides a means to analyze a temporal data at multiple resolutions and thus the clusters are obtained at multiple resolutions. The techniques have been tested on an EEG dataset recorded through 64 channels