Anomaly Detection in Large-Scale Trajectories Using Hybrid Grid-Based Hierarchical Clustering

Abstract

The increasing availability of location-acquisition technologies (such as GPS and GSM networks) and mobile computing techniques has generated a lot of spatial-temporal trajectory data and indicates the mobility of diversified moving objects such as people, vehicles, and animals. This brings new opportunities to identify abnormal activities of moving objects. This paper describes our detection of anomalies in human trajectory data using a hybrid grid-based hierarchical clustering method based on Hausdorff distance, which is suitable for measuring the similarity between trajectories of different lengths. The trajectories were first transformed into grid-based trajectories using a grid structure. After that, the grid-based trajectories were clustered based on their pairwise Hausdorff distances by applying different versions of hierarchical clustering algorithms. We evaluated our research result using a reallife dataset (published by Microsoft Research Asia), ground truth reconstructed by us, and evaluation criteria widely used in data mining. The experimental results demonstrate that the proposed algorithm is more effective and much faster than the traditional hierarchical clustering algorithm according to the pairwise comparison results.

A Review of Computational Methods for Clustering Genes with Similar Biological Functions

“…This method performs faster than density-based clustering. Grid-based clustering can benefit from dividing the data space into grids to reduce its time complexity [22,60]. CLIQUE, grid-clustering technique for high-dimensional very large spatial databases (GCHL), and statistical information grid (STING) are examples of grid-based clustering [39][40][41][42][43].…”

Explainable Anomaly Detection Framework for Maritime Main Engine Sensor Data

“…Then, the closest pair of clusters is merged based on a distance measure given by the user. This is repeated until all the points are merged into one root cluster [ 49 ]. As shown in Figure 6 , as the data is grouped and groups are merged, the cluster tree represented in the dendrogram was formed.…”

Bioinformatics Applications Based On Machine Learning

“…This method performs faster than density-based clustering. Grid-based clustering can benefit from dividing the data space into grids to reduce its time complexity [22,60]. CLIQUE, grid-clustering technique for high-dimensional very large spatial databases (GCHL), and statistical information grid (STING) are examples of grid-based clustering [39][40][41][42][43].…”

“…':0.1340, 'T':0.1335, 'A':0.1260, 'G':0.0806} {'C':−1, 'T':−2, 'A':2, 'G'':0.11, 'TT':0.091, 'AA':0.091, 'GG':0.11, 'CT':0.078, 'TA':0.06, 'AG':0.078, 'CA':0.058, 'TG':0.058, 'CG':0.119, 'TC':0.056, 'AT':0.086, 'GA':0.056, 'AC':0.065, 'GT':0.065, 'GC':0.1111}[60] …”