Using Multilayer Perceptron in Computer Security to Improve Intrusion Detection

摘要: Nowadays computer and network security has become a major cause of concern for experts community, due to the growing number devices connected network. For this reason, optimizing performance systems able detect intrusions (IDS - Intrusion Detection System) is goal common interest. This paper presents methodology classify hacking attacks taking advantage generalization property neural networks. In particular, in work we adopt multilayer perceptron (MLP) model with back-propagation algorithm sigmoidal activation function. We analyse results obtained using different configurations network, varying hidden layers training epochs order obtain low false positives. The will be presented terms type show that best classification carried out DOS Probe attacks.