Project Overview: Agent-Based Learning to Utilize Local Data for Anomalous Activity Recognition

Learn more about Agent-Based Learning to Utilize Local Data for Anomalous Activity Recognition from M. Hadi Amini, Assistant Professor at Florida International University.

In the mission to protect the nation from ever-evolving threats, DHS requires an automated way to detect anomalous activity in large amounts of video data, and share this information securely across organizational boundaries without compromising privacy. This project’s proposed approach will generate a cumulative, agent-based machine learning model to detect suspicious activity and improve detection accuracy across video sources, without the need for sensitive video data to be shared between sites. The system and user-friendly interface developed in this project can be integrated into existing systems of stakeholders such as USSS, USCIS, USCG, CBP, TSA and ICE, allowing local video to be processed onsite and prioritized for review by a human analyst more efficiently and effectively.