The Intrusion Detection System (IDS) enforces information security and is responsible to identify attacks and vulnerabilities inside a network. It does this by analyzing the packet stream throughout the network. In traditional IDS systems, the analysis is done by looking for signatures of known attacks or deviations of normal activity as described by the rules provided for the IDS system. Machine learning helps in deriving predictive knowledge and this makes it ideal to apply Machine learning in an IDS system to detect attacks. This paper focuses on creating a hybrid model that is best to implement in an IDS system. A hybrid model is implemented which combine multiple machine learning algorithms using Ensemble method. The experiments include evaluating machine learning algorithms such as Decision Tree, MLP (Multi-Layer Perceptron), Gradient Boosting etc. The algorithms with the best results are taken to construct Hybrid model. This Hybrid approach will improve the accuracy and efficiency for identifying the attacks by the IDS system. Depending on the type of attack, the IDS system can classify packets as DoS (Denial of Service), Probe, R2L (Root to Local), U2R (User to Root) or Normal. The experiments are carried using NSL-KDD Dataset.
Volume 11 | 06-Special Issue