Similarity search in seismological signals

Similarity search in time series databases has been the focus of a lot research. It is non-trivial problem due their peculiar structure, high dimensionality (C. Faloutsos et. al, 1994), noise, and high feature correlation; it is also the heart of many data mining applications, like clustering, classification, rule discovery and query by content. In general, similarity has been measured with the Euclidean metric, but the dimensionality curse problem remains (K. Chan and A. Wai-chee, 1999). Basically, there are four major techniques that have been proposed in order to reduce the dimensionality of the data (E. Keogh et. al, 2000), singular value decomposition (SVD), the discrete Fourier transform (DFT), the discrete wavelets transform (DWT), and the piecewise aggregate approximation (PAA)(E. Keogh and M. Pazzami, 2000). In this work, we use the dynamic time warping (DTW) algorithm and PAA dimensionality reduction technique as a new approach to find the similarity between seismological signals.