New Intrusion Detection System Based on Neural Networks and Clustering

Abstract

Efficiency of Intrusion detection systems-IDS are evaluated using parameters like completeness, performance and accuracy. The first important parameter is the completeness, which occurs when the detection of attack fails. This is the most difficult parameter to evaluate compared to the other two parameters. The second one is performance, which indicates the audit events process. When the IDS doesn’t work properly or works poorly, the real time detection becomes impossible. Legitimate actions are flagged as anomalous which is termed as inaccuracy. This part needs attention to address the inaccuracies. Optimal solutions must take the inaccuracies into consideration for accuracy, thereby efficiency of IDS. There are different trends in IDS. Some of them are discussed below. Behavior and knowledge-based IDS: Misuse detection, appearance-based detection, behavior detection and anomaly detection etc. There are numerous stability and security issues as a result of the Internet’s and computer networks’ rapid proliferation. The present study reports the case study of image processing in a fruit grading plant with data safety over cloud with Original Equipment Manufacturer (OEM). How Artificial Neural Networks (ANN) architecture can help is discussed and recommendations are made for impending improvement.