Group Feature Selection via Structural Sparse Logistic Regression for IDS

Intrusion Detection System (IDS) may be considered one of the most important components in the domain of network security. Currently, most of the IDSs encounter some problems like feature redundancy, high-dimensional feature, overfitting and a limited number of training examples. Thus, feature selection acts as an important role in IDSs to solve the aforementioned problems. In our suggested approach, we try to address these problems via Structural Sparse Logistic Regression (SSPLR). In this paper, we propose important feature groups and individual feature selection as well as intrusions classification for IDSs. Recently, SSPLR has been used as a technique for data analysis and processing via structural sparse penalization. With SSPLR, the relation between features has been considered in the modeling. Finally, the experiments in this correspondence demonstrate that the SSPLR technique has better performance than the currently used techniques addressing the same challenge.