[2M20-22] Nuclear Non-proliferation and Nuclear Security 2

[2M22] Detection and Prediction of Insider Malicious Behavior for Nuclear Security using Kinect

*Shi Chen¹, Kazuyuki Demachi¹, Tomoyuki Fujita¹, Yutaro Nakashima¹, Yusuke Kawasaki¹ (1.The University of Tokyo)

Keywords:Malicious Behavior Detection, Hand Motion Tracking, Kinect, Auto-Associate Neural Network

After Fukushima Daiichi accident, the importance of nuclear security is increased. Especially as the threat to nuclear power plants, sabotage by insider is significant.
In this research, the technology to detect malicious behaviors by insiders was developed by using Microsoft Kinect v2. Time series data analysis is a useful method in their abnormal behavior detection. However, malicious behavior detection isn’t enough for nuclear security, future malicious behavior prediction should be taking into consideration.