Ensemble of Soft Computing Techniques for Intrusion Detection
In the present world scenario network-based computer systems have started to play progressively more vital roles. As a result they have become the main targets of our adversaries. To apply high security against intrusions and attacks, a number of software tools are being currently developed. To solve the problem of intrusion detection a number of pattern recognition and machine learning algorithms has been proposed. The paper states the problem of classifier fusion with soft labels for Intrusion Detection. Performance of Artificial Neural Networks (ANN) and Support Vector Machines (SVM) is presented here. The performance of fusing these classifiers using approaches based on Dempster- Shafer Theory, Average Bayes Combination and Neural Network is proposed. As shown through the experimental results combined classifiers perform better than the individual classifiers.
