Using Fuzzy Neural Networks and rule heuristics for anomaly intrusion detection on database connection
Rung-Ching Chen, Kai-Fang Cheng, Cheng-Chia Hsieh · 2008
This paper addresses the issue of intrusion detection in database security management. A Fuzzy Adaptive Resonance Theory neural network and rule heuristics are used to build a model of company security judgment. The model is based on analysis of the log file of connections from the client side to the database of server side. The log file information includes user name, network address of client, the time of connection, the database name, the program used, and the protocol. Those features are inputted to a Fuzzy Adaptive Resonance Theory neural network for security judgment. An experiment using records from a local government office database indicates that our system has good results in detecting anomalous intrusions.