The study on the abnormal intrusion detection performances based on artificial neural networks

• Main Authors: Lee,Chang yuan, 李鎮原
• Other Authors: BRANDT TSO
• Format: Others
• Language: zh-TW
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/60713265655259148711

碩士 === 國防管理學院 === 資源管理研究所 === 93 === In recent years, the fast development of internet not only provides the convenient ways of acquiring the information, but also let the hackers can use internet as media to conduct intrusions and corruptions. This may cause the victim considerable damages. Therefore, the internet security has been deeply appreciated. In the field of network security, the intrusion detection is the one of most important domain because such detection measure could be the last wall of the defense. If there is a reliable system as our backup, one can then soon realize the situation being facing and the condition of the environments in where the system locates. Also, whenever there is an intrusion happening, the system may quickly receive the warning and take relevant countermeasure. Normally, there are two kinds of factors may affect the efficiencies of intrusion detection system: one is the level of analysis tools, the other is the detection factors for determining intrusion underway. In this study, we focus on analyzing the efficiency of intrusion detection analysis tool. The relevant tools under analysis include perceptron networks, back-propagation networks, probability networks, and regression networks, so as to investigate the different neural networks performances and responses to abnormal detection within networks.