Neural Networks and Decision Trees for Intrusion Detections: Enhancing Detection Accuracy

EasyChair Preprint no. 2897

Abstract

Many artificial intelligence methods were applied to enhance security in computer networks. These methods seem to be mainly based on neural networks and decision trees. Nevertheless, and according to literature, some of them are still suffering from some weaknesses. This is the reason why we focused in this study on the enhancement of two approaches: Iterative Dichotomiser 3 ID3 and multilayer perceptron (MLP) algorithms. The aim of this study is to use appropriate attributes of the KDD dataset, in order to obtain better detection rates. Simulations were conducted using WEKA and Tanagra tools. The results show that our contributions (ID3 and MLP) are competitive with other solutions in terms of detection accuracy.

Keyphrases: 10%KDD cup’99 Dataset, decision trees, experimental comparison, ID3, IDS, MLP Neural Network

@Booklet{EasyChair:2897,
  author = {Rachid Beghdad and Katia Bechar and Meriem Bouali and Haddadi Mohamed},
  title = {Neural Networks and Decision Trees for Intrusion Detections: Enhancing Detection Accuracy},
  howpublished = {EasyChair Preprint no. 2897},

  year = {EasyChair, 2020}}