Abstract
Understanding of human dynamics has drawn attention to various areas. The wide spread of positioning technologies using GPS facilitates location information to be obtained with high spatial-temporal resolution as well as at low cost. By collecting individual location information in real time, monitoring of human dynamics is recently considered possible and is expected to expand the area of dynamic traffic control in the future. In this monitoring, detecting anomalous states in human dynamics becomes important. This research aims to define an anomaly detection problem of the human dynamics monitoring with gridded population data and propose an anomaly detection method based on the definition. First we discussed the characteristics of the anomaly detection in human dynamics monitoring and categorized our problem to a semi-supervised anomaly detection problem that detects contextual anomalies behind time-series data. We proposed an anomaly detection method based on a sticky hierarchical Dirichlet process hidden Markov model, which is able to estimate the number of latent states according to the input time-series data. Results of the experiment with synthetic data showed that our proposed method, which detects anomalies by comparing posteriors of latent state variable, has good performance with respect to both detection rate and precision. Through the experiment with real gridded population data, high anomaly scores were detected at the time when train services had been stopped.
