Design and Implementation of a Hybrid Deep Learning Algorithm for IoT Attack Detection Using Harmony Search for Hyperparameter Optimization and Feature Selection
Keywords:
Harmony search algorithm, Neural Networks, CNN-LSTM model, feature selectionAbstract
Objective: The primary objective of this study is to design and implement a hybrid deep learning algorithm that integrates CNN and LSTM architectures for the proactive detection of attacks in IoT networks. Methodology: This study develops a hybrid CNN-LSTM algorithm for IoT attack detection, utilizing Harmony Search for both hyperparameter optimization and feature selection. The Harmony Search algorithm is applied in two optimization stages: first for tuning hyperparameters, and second for selecting relevant features. The hybrid model is trained on IoT data and evaluated based on metrics such as accuracy, precision, recall, and F1 score. Findings: The results indicate that the CNN-LSTM model optimized with Harmony Search demonstrates superior accuracy and efficiency in detecting IoT attacks compared to traditional methods, leading to improved key performance indicators and reduced computational overhead. Conclusion: This approach can serve as a robust solution for proactive IoT attack detection and other security challenges. Objective: The primary objective of this study is to design and implement a hybrid deep learning algorithm that integrates CNN and LSTM architectures for the proactive detection of attacks in IoT networks. Methodology: This study develops a hybrid CNN-LSTM algorithm for IoT attack detection, utilizing Harmony Search for both hyperparameter optimization and feature selection. The Harmony Search algorithm is applied in two optimization stages: first for tuning hyperparameters, and second for selecting relevant features. The hybrid model is trained on IoT data and evaluated based on metrics such as accuracy, precision, recall, and F1 score. Findings: The results indicate that the CNN-LSTM model optimized with Harmony Search demonstrates superior accuracy and efficiency in detecting IoT attacks compared to traditional methods, leading to improved key performance indicators and reduced computational overhead. Conclusion: This approach can serve as a robust solution for proactive IoT attack detection and other security challenges.