Hybrid optimization based deep stacked autoencoder for routing and intrusion detection

Article type: Research Article

Authors: Boopathi, Mythili^{; *}

Affiliations: School of Computer Science Engineering and Information Systems, Vellore Institute of Technology, Katpadi, Vellore-632014, Tamil Nadu, India

Correspondence: [*] Corresponding author. E-mail: [email protected].

Abstract: This research introduced the optimized Deep Stacked Autoencoder (DSA) for performing Intrusion Detection (ID) in the IoT. Firstly, IoT simulation is carried out and then, the information is routed by using the Chronological War Strategy Optimization (CWSO). Here, the CWSO is newly designed by incorporating the chronological concept with the WSO. After the routing, the ID is completed at the Base station (BS) by executing the following steps. Initially, data is obtained from a database, after that, feature normalization is done using min-max normalization. Meanwhile, Canberra distance is applied to execute the feature selection process. Finally, ID is performed using DSA, which is trained using the Competitive Swarm Henry War Strategy Optimization algorithm (CSHWO). The experimental result confirms that the invented scheme accomplished the superior outcome by the energy, f-score, precision, and recall values of 0.379, 0.913, 0.918 and 0.912, respectively.

Keywords: Competitive Swarm Optimization, War strategy Optimization, Henry gas solubility optimization, Deep Stacked Autoencoder, Chronological concept

DOI: 10.3233/WEB-230109A

Journal: Web Intelligence, vol. Pre-press, no. Pre-press, pp. 1-20, 2024