International Conference on Computer and Knowledge Engineering

Home / 15th International Conference on Computer and Knowledge Engineering

An Evolutionary Approach with Surrogate Models for Feature Selection in Intrusion Detection Systems

Authors :

Sadeq Moradi¹ Hadi Shahriar Shahhoseini²

1- School of Electrical Engineering, Iran University of Science and Technology 2- School of Electrical Engineering, Iran University of Science and Technology

Keywords :

intrusion detection systems،feature selection،evolutionary algorithms

Abstract :

With the fast progress of technology, network traffic volume has grown significantly. This has caused the emergence of datasets with a large number of features. These features, due to both noise and high dimensionality, reduce the performance of Intrusion Detection Systems (IDS). In this paper, an evolutionary algorithm is proposed that employs a population grouping method to increase the diversity of solutions. A surrogate model, which is managed using the population grouping method, and local search is then integrated into the algorithm to design a feature selection framework. The algorithm's performance is evaluated using standard CEC benchmark functions, and the results demonstrate its comparable capability in finding the optimal solution. Furthermore, the proposed feature selection framework is evaluated on two datasets. The results indicate that this method performs well compared to common algorithms, achieving a Recall of 91.30% on the NSL-KDD dataset and 99.65% on the UNSW-NB15 dataset.