核セキュリティのための内部脅威者の自動検知技術の開発
Automatic Detection of Malicious Insider Behavior for Nuclear Security
著者:
川崎 祐典 Yusuke KAWASAKI 出町 和之 Kazuyuki DEMACHI 笠原 直人 Naoto KASAHARA
発刊日:
公開日:
概要
Abstract After Fukushima accident, the threats of terrorism are increasing. However, countermeasures for insider terrorists are insufficient because the main target of Physical Protection System (PPS) is outsider. Thus development of automatic detection of insider terrorists is required. In this paper, focusing on the hand motion, final goal is image recognition by Convolutional Neural Network (CNN) and analysis of each finger’s position and angle and detection of signs of malicious behavior. And For this, results of trying CNN and making dataset for hand motion recognition are showed. Future work is development of analysis method of time-series feature data and detection of malicious behavior. Keywords: Nuclear security, Insider terrorists, Abnormal detection, Image analysis, Image recognition
