A Novel Ensemble Approach for Effective Intrusion Detection System

Machine learning algorithms are useful for decision making on valuable datasets which are using in emerging fields such as networks, medical and e-governance. Ensemble classifier is a most useful approach which is the combination of classification algorithms for performing effective classification in machine learning. Even though, the selection of ensemble is becoming very difficult task for the specific dataset. For this purpose, we introduce a novel ensemble which is the combination of attribute selection algorithm, multiclass support vector machine and k-NN classifier. Moreover, we use an Incremental Particle Swarm Optimization (IPSO) for optimizing the proposed system performance in terms of improving the classification accuracy. For conducting various experiments, we have used the five random subsets from the standard KDD'99 dataset.