علي اسماعيل

Abstract Cloud computing, Internet of Things (IoT), an‎d large-scale digital infrastructures have grown so quickly that the complexity an‎d frequency of cyber-attack threats have multiplied in about the same order. In general, traditional intrusion detection system (IDS) has difficulty achieving high detection accuracy an‎d low false positive rate in heterogeneous an‎d large-scale environments. In response to these challenges, this thesis offers a hybrid deep learning-based intrusion detection framework that combines Long Short-Term Memory (LSTM) networks with the Chimp Optimization Algorithm (ChOA) for hyperparameter tuning in an adaptive manner. The designed LSTM–ChOA model exploits temporal sequence learning to represent changing traffic dynamic behavior alongside metaheuristic optimization for stability of convergence an‎d overall performance. The performance of the framework has been assessed on two benchmark datasets (CICIDS2017 an‎d UNSW-NB15) in binary as well as multi-class classification settings. Experimental results show high detection accuracy an‎d improvements of F1-scores an‎d false positive rates with respect to baseline models. Multiple runs of each experiment were performed to ensure robustness an‎d all the results were statistically validated confirming consensus. Domain Shift (Cross-dataset) Experiments: Further validating the generalization capability of the model under domain shift. Moreover, GPUs showed scalability an‎d real-time detection capability in big-data environment based on distributed training. In conclusion, the introduced LSTM–ChOA framework introduces a strong, scalable an‎d high-performance means of modern intrusion detection for dynamic networks. Keywords: Intrusion Detection System, LSTM, Chimp Optimization Algorithm, Deep Learning, Big Data Analytics, Cross-Dataset Generalization, Network Security.

سيستم تشخيص نفوذ , LSTM , الگوريتم بهينه‌سازي شامپانزه , يادگيري عميق , ، تحليل كلان‌داده , تعميم بين‌مجموعه‌داده‌اي , امنيت شبكه

Intrusion Detection System , LSTM , Chimp Optimization Algorithm , Deep Learning , Big Data Analytics , Cross-Dataset Generalization , Network Security.

Ali Ismaeel

Dr. Hassan Naderi