Paper Contents
Abstract
IDS systems face challenges such as large feature sets, imbalanced datasets, and a high number of samples, which can affect performance. To address these challenges, this research proposes an SVM-based IDS system that utilizes Fisher's score for feature selection, over- and under-sampling techniques for balancing the dataset, and a probabilistic sample reduction technique to reduce the size of the dataset. Fisher's score is applied to select the most relevant features from a large feature set, while over- and under-sampling techniques balance the dataset by over-sampling the minority class and under-sampling the majority class. Additionally, a probabilistic training data sampling technique is employed to reduce the size of the dataset and, hence, training time and resource requirements. The proposed SVM-based IDS system is evaluated on the KDD99 dataset, and the experimental results demonstrate its superiority over existing SVM-based IDS systems in terms of accuracy, precision, recall, and F1-score
Copyright
Copyright © 2023 Sheetal Kori. This is an open access article distributed under the Creative Commons Attribution License.
