A Hybrid Approach for Intrusion Detection System to Enhance Feature Selection

Finding the most important variables for a given job is about feature selection, a key technique in machine learning and data analysis, entails. In the creation of machine learning models, it is essential. The dataset’s irrelevant characteristics might have a negative impact on the model’s accuracy and training time. For the construction of an Intrusion Detection System (IDS), feature selection holds significant importance. With the increasing scale of data, feature selection becomes even more important to improve the accuracy and efficiency of machine learning models and avoid overfitting. This study provides an overview of the importance of feature selection and introduces common techniques specifically designed for IDS. While a hybrid method based on factor analysis, mRMR, SFS, and FSF-score has shown promising results in accuracy and dimensionality reduction, it has limitations such as the accuracy of factor analysis in datasets with non-linear relationships and the assumption that the maximum load is always the most informative aspect of a feature. Therefore, it is crucial to carefully consider the limitations and drawbacks of this approach for different datasets and classification tasks. To overcome these limitations, the Boruta feature selection algorithm is proposed as an alternative solution.