The Role of Neural Networks in Safeguards and Security

Interpreting data from nuclear safeguards and computer security systems is a tedious and time-consuming task. It typically requires the examination of large amounts of data for unusual patterns of activity. Neural networks provide a flexible pattern-recognition capability that can be adapted for these purposes. In this paper we describe a methodology for performing anomaly detection and consistency checking in safeguards and security data.