A Copula Entropy‐Assisted Hierarchical Binary Swarm Intelligence Algorithm‐Enabled Feature Selection and Intrusion Detection for IoT Networks

ABSTRACT Feature selection (FS) is a critical preprocessing approach for enhancing the quality of feature sets in various real‐world applications. Due to diverse relationships among features in high‐dimensional data sets, various features exhibit differing levels of importance and correlation with the target, which makes it challenging to effectively filter out irrelevant or redundant features. In this paper, we present a copula entropy‐assisted hierarchical binary swarm intelligence optimiser to address high‐dimensional FS problems. Specifically, the dung beetle optimiser is suitable for solving FS problems by designing binary state labels and constructing a hierarchical mechanism. Then, to alleviate the issue of low‐quality initial solutions, the copula entropy is incorporated into the initial population, which can enhance the diversity of solutions and provide more robust starting points. Next, the effectiveness of the developed algorithm is tested on different high‐dimensional data sets. Experimental results demonstrate that the proposed method surpasses other state‐of‐the‐art swarm intelligence‐based FS approaches. Finally, our proposed algorithm is successfully applied in the intelligent intrusion detection systems for Internet of Things networks.