Fast retrieval of time series using a multi-resolution filter with multiple reduced spaces

Abstract

Fast retrieval of time series that are similar to a given pattern in large databases is a problem which has received a lot of attention in the last decade. The high dimensionality and large size of time series databases make sequential scanning inefficient to handle the similarity search problem. Several dimensionality reduction techniques have been proposed to reduce the complexity of the similarity search. Multi-resolution techniques are methods that speed-up the similarity search problem by economizing distance computations. In this paper we revisit two of previously proposed methods and present an improved algorithm that combine the advantages of these two methods. We conduct extensive experiments that show the show the superior performance of the improved algorithm over the previously proposed techniques. © 2010 Springer-Verlag.