Comparative survey on SVM kernel’s performance for classification used in IDS

Our work covers implementation of an IDS using Machine Learning. For the experiment purpose we are using KDDCUP’99 dataset. Even after having few demerits as covered in this paper, KDDCUP’99 dataset is widely used for network security related research worldwide. For the comparison purpose we are using 10% of the dataset. Both the full dataset and a 10% of it are globally available in the internet. As a part of Machine Learning, hear we are using SVM for classification of the intrusion type. SVM is considered as one of the best models used for classification task. Depending upon complexity of the dataset, there are various selected kernels working as part of SVM model. These kernels are used to best fit data in the model and enhance its performance. While implementing an IDS, our work focuses on a comparison among few of these kernels in terms of accuracy.