2026-10-12T04:32:32Z https://api.figshare.com/v2/oai

oai:figshare.com:article/34035779 2026-09-30T17:40:29Z category_146 category_734 category_931 portal_5 item_type_3 month_year_09_2026

<p>Ablation study of the proposed framework.</p> Jisi Chandroth (25145375) Gabriel Stoian (5712970) D.Jude Hemanth (25145378) Space Science Biological Sciences not elsewhere classified Information Systems not elsewhere classified spread risk throughout smart home appliances single vulnerable node scale security concerns jeopardize user privacy improve discriminative learning generation wireless networks experimental results show enables intelligent services bidirectional long short becoming increasingly interconnected ant colony optimization proposed model draws proposed model combines accurately recognize threats intrusion detection systems highly accurate attack smart ce network intrusion detection proposed framework attack surface detection solution detecting threats cyber threats uncorrelated attributes term memory recent years nb15 dataset important representations extract informative exposing resource evaluated using cloud infrastructures attention layer also widens <div><p>Consumer Electronics (CE) devices, such as smartwatches, cameras, and smart home appliances, are becoming increasingly interconnected through Smart CE networks supported by Internet of Things (IoT) ecosystems and next-generation wireless networks. This ubiquitous connectivity enhances user convenience and enables intelligent services, but it also widens the attack surface by exposing resource-constrained devices to cyber threats. Although many CE devices constantly communicate with edge or cloud infrastructures, compromising a single vulnerable node can spread risk throughout the Smart CE network and jeopardize user privacy. Intrusion Detection Systems (IDS) are commonly used security systems for detecting threats and vulnerabilities in consumer devices. Although several IDS techniques have been developed in recent years, the Smart CE network environment still requires a real-time, highly accurate attack-detection solution to address its ever-changing, large-scale security concerns. In this paper, we propose a hybrid intrusion detection framework for securing smart CE network. The proposed model draws on the strengths and capabilities of multiple deep learning algorithms. Specifically, the proposed model combines a Deep Convolutional Neural Network (DCNN) and a Bidirectional Long Short-Term Memory (BiLSTM) network to accurately recognize threats. In addition, we use the Ant Colony Optimization (ACO) approach to extract informative and uncorrelated attributes. An attention layer is added to improve discriminative learning by emphasizing the most important representations. The proposed framework was evaluated using the UNSW-NB15 dataset. The experimental results show that the proposed framework is more accurate than the existing techniques.</p></div> 2026-09-30T17:40:15Z Dataset Dataset 10.1371/journal.pone.0342949.t005 https://figshare.com/articles/dataset/_p_Ablation_study_of_the_proposed_framework_p_/34035779 CC BY 4.0