Abstract:Urban anomalies, such as abnormal movements of crowds and accidents, may result in loss of life or property if not handled properly. It would be of great value for governments if anomalies can be automatically alerted in their early stage. However, detecting anomalies in urban area has two main challenges. First, the criteria to determine an anomaly on different occasions (e.g. rainy days vs. sunny days, or holidays vs. workdays) and in different places (e.g. tourist attractions vs. office areas) are distinctly different, as these occasions and places have their own definitions on normal patterns. Second, urban anomalies often exhibit complex forms (e.g. road closure may cause decrease in taxi flow and increase in bike flow). We need an algorithm that not only models the anomaly degree of individual data source but also the combination of changes in multiple data sources. In this paper, we propose a two-step method to tackle those challenges. In the first step, we use a similarity-based algorithm to estimate an anomaly score for each individual data source in each region and time slot based on the values of historically similar regions. Those scores are fed into the second step, where we propose an algorithm based on one-class Support Vector Machine to capture rare patterns occurred in multiple data sources, nearby regions or time slots, and give a final, integrated anomaly score for each region. Evaluations based on both synthetic and real world datasets show the advantages of our method beyond baseline techniques such as distance-based, probability-based methods.

AnomalyDetect: An Online Distance-Based Anomaly Detection Algorithm

Anomaly Detection Algorithm Based on Electric Equipment

An Unsupervised Anomaly Detection Algorithm for Time Series Big Data

Interactive Context-Aware Anomaly Detection Guided by User Feedback

Distributed Online One-Class Support Vector Machine for Anomaly Detection over Networks

Anomaly Detection Model Based on Hadoop Platform and Weka Interface

A simple method for unsupervised anomaly detection: An application to Web time series data

An Experimental Evaluation of Anomaly Detection in Time Series

Exact variable-length anomaly detection algorithm for univariate and multivariate time series

An Empirical Analysis of Anomaly Detection Methods for Multivariate Time Series

The Design And Implementation Of A Multidimensional And Hierarchical Web Anomaly Detection System

An Extreme Learning Machine for Unsupervised Online Anomaly Detection in Multivariate Time Series

Detecting Urban Anomalies Using Multiple Spatio-Temporal Data Sources

Anomaly detection in online social networks

Online Model-based Anomaly Detection in Multivariate Time Series: Taxonomy, Survey, Research Challenges and Future Directions

Anomaly Detection at Scale: The Case for Deep Distributional Time Series Models

A Comparative Study on Unsupervised Anomaly Detection for Time Series: Experiments and Analysis

A Time-Stamp Frequent Pattern-Based Clustering Method For Anomaly Detection

Unsupervised Learning Based Distributed Detection of Global Anomalies.

The Financial Data of Anomaly Detection Research Based on Time Series

HybridAD: A Hybrid Model-Driven Anomaly Detection Approach for Multivariate Time Series