Machine Learning for Identifying Group Trajectory Outliers

Prior works on the trajectory outlier detection problem solely consider individual outliers. However, in real-world scenarios, outliers can often appear groups, e.g., a group of bikes that deviates to usual due maintenance streets context intelligent transportation. The current paper considers Group Trajectory Outlier (GTO) and proposes three algorithms. first second algorithms are extensions well-known DBSCAN k NN algorithms, while third one models GTO as feature selection problem. Furthermore, two different enhancements for proposed proposed. is based ensemble learning computational intelligence, which allows merging algorithms’ outputs possibly improve final result. general high-performance computing framework deals with big databases, we used GPU-based implementation. Experimental results real databases show scalability approaches.