XaIBO: An extension of aIB for trajectory clustering with outlier

Abstract

Clustering plays an important role for trajectory analysis. The agglomerative Information Bottleneck (aIB) approach is effective for successfully managing an optimum number of clusters without the need of an explicit measure of trajectory distance, which is usually very difficult to be defined. In this paper, we propose to utilize a statistically representation of the trajectory shape to perform the aIB based trajectory clustering. In addition, an extension of aIB is proposed to cope with the clustering on trajectories with outliers (for brevity, we call this extended version of aIB as XaIBO) and in this case, XaIBO can be widely used in practice for dealing with complex trajectory data. Extensive experiments on synthetic, simulated and real trajectory data have shown that XaIBO achieves the trajectory clustering very well.