Intrusion Detection Method Using Neural Networks Based on the Reduction of Characteristics

摘要: The application of techniques based on Artificial Intelligence for intrusion detection systems (IDS), mostly, artificial neural networks (ANN), is becoming a mainstream as well an extremely effective approach to address some the current problems in this area. Nevertheless, selection criteria features be used inputs ANNs remains problematic issue, which can put, nutshell, follows: wider spectrum selected is, lower performance efficiency process becomes and vice versa. This paper proposes sort compromise between both ends scale: model Principal Component Analysis (PCA) chosen algorithm reducing characteristics order maintain without hindering capacity detection. PCA uses data diminish size ANN's input vectors, ensuring minimum loss information, consequently complexity classifier maintaining stability training times. A test scenario validation purposes was developed, using based-on-ANN IDS. results obtained tests have demonstrated validity proposal.