Extending the Gaussian membership function for finding similarity between temporal patterns

Abstract

In this paper, the basic Gaussian membership function is extended to design the dissimilarity measure. We extend the dissimilarity measure proposed in the G-spamine by applying normal distribution. The dissimilarity measure proposed in this paper is designed by using the concept of standard normal distribution. For a pattern to be similar, the dissimilarity between reference and the temporal pattern has to be less than or equal to the dissimilarity constraint. This dissimilarity constraint is obtained by transforming the user threshold value to z-space. The dissimilarity measure has also been extended to compute the distance bounds by devising necessary expressions. These distance bounds can be used to prune the invalid temporal associations. The algorithm to obtain the similar temporal associations is outlined.