New Model To Detect Anomalies in Human Mobility Considered Collective Behavior, Outperformed Existing Approaches

Detecting anomalies in human mobility is essential for applications such as public safety and urban planning. Traditional methods of detection focus on individual patterns of movement and identify deviations from routine behaviors, but people often interact with others, forming collective behaviors. These collective behaviors reveal anomalies driven by human interactions, which have largely been ignored in research.

In a new study, researchers proposed a new model designed to detect anomalies based on individuals’ collective behaviors. Their model significantly outperformed existing approaches to detection.

The study, by researchers at Carnegie Mellon University, appears in ACM Digital Library.

“Although researchers have extensively studied human mobility modeling and anomaly detection, none of the existing approaches effectively addresses the many challenges inherent in this work,” says Leman Akoglu, associate professor of information systems at Carnegie Mellon’s Heinz College, who coauthored the study.

Unlike individual anomalies, collective anomalies require modeling the spatio-temporal dependencies between individuals over time, adding a layer of complexity to research on this issue. In this study, researchers developed CoBAD, the first model designed to capture collective behaviors for detecting anomalies in human mobility.

CoBAD uses a two-stage attention mechanism to model both individual mobility patterns and interactions among multiple individuals. The model is trained to reconstruct both the attributes of randomly masked events (capturing individual behaviors) and their associated links (capturing collective behaviors).

Based on experiments on large-scale mobility datasets, the study found that CoBAD significantly outperformed existing anomaly detection baselines by 13% to 70%.

“By formally defining related concepts and presenting problem formulation for detecting collective anomalies in human mobility, we were able to detect patterns based on both attributes of events and their co-occurrence relations,” explains Haomin Wen, a postdoctoral student at Carnegie Mellon’s Heinz College, who led the study.

“We expect our work and open-source codebase to foster further research on modeling human mobility through the lens of social interactions and collective dynamics,” adds Shurui Cao, a PhD student at Carnegie Mellon’s Heinz College, who coauthored the study.

The study was supported by the Intelligence Advanced Research Projects Activity of the U.S. Department of Interior/Interior Business Center.

###

Summarized from an article in ACM Digital Library, "CoBAD: Modeling Collective Behaviors for Human Mobility Anomaly Detection," by Wen, H (Carnegie Mellon University), Cao, S (Carnegie Mellon University), and Akoglu, L (Carnegie Mellon University). Copyright 2025 The Authors. All rights reserved

About Carnegie Mellon University's Heinz College of Information Systems and Public Policy

The Heinz College of Information Systems and Public Policy is home to two internationally recognized schools: the School of Information Systems and Management and the School of Public Policy and Management. Heinz College leads at the intersection of people, policy, and technology, with expertise in analytics, artificial intelligence, arts & entertainment, cybersecurity, health care, and public policy. The college offers top-ranked undergraduate, graduate, and executive education certificates in these areas. Our programs are ranked #1 in Information Systems, #1 in Information and Technology Management, #8 in Public Policy Analysis, and #1 in Cybersecurity by U.S. News & World Report. For more information, visit www.heinz.cmu.edu.