Spatiotemporal Clustering from Noisy Data

P. Kanjilal, S. Das, and D. Lawless

IJCAI Workshop on Spatial and Temporal Reasoning, Hyderabad, India (January?, 2007)

Spatiotemporal clustering of multiple objects in a simulated space is considered. New preprocessing methods using Singular Value Decomposition (SVD) and Approximate Entropy are discussed to reduce clutter. A systematic procedure is presented to identify spatiotemporal clusters in terms of optimal sets of singular vectors. A novel orthogonal transformation based approach is proposed for initializing the k-mean square clusters.

Request a Copy