Intrusion Detection using HRO with Ensemble Learning Models and Comparison

An ancient form of security technology is intrusion detection, consisting of a set of programs that observes traffic of network and gives a warning message when a cautious thing takes place. Hence four intrusion detection techniques are considered. For Support Vector Machine radial kernel function, decision trees for Random Forest and the Extreme Learning Machine, a single layer feed-forward neural network likewise the specific model intrusion detection system implementation has been carried out and the result is that ELM gives accurate result in accuracy. ELM is examined further to study its performance through optimizing parameters with the help of Hybrid rice algorithm. Mainly these methods are parameter dependent and in regardance with performance to a great extent. In this, proposed system the extreme learning machine parameters are encoded as rice gene location, Then the optimal parameters are found by simulating rice breeding behavior then the result of test accuracy fitness function represents the HRO optimized ELM algorithm gives more accuracy than others, then precision and recall values are also measured. Through simulation of rice breeding the ELM best parameters are noted, concludes that HRO based ELM improves the intrusion detection accuracy.