Intelligent Threat Detection in Cloud Computing Using Machine Learning Techniques: A Hybrid Intrusion Detection Framework

Cloud computing has fundamentally transformed how organizations deploy, scale, and manage computational resources. However, this paradigm shift introduces significant security vulnerabilities arising from multi-tenancy, distributed architectures, and dynamic traffic patterns. Traditional intrusion detection systems struggle to address these challenges due to their reliance on static signature-based methods and inability to adapt to novel attack vectors. This paper presents a hybrid intrusion detection framework that integrates machine learning and deep learning techniques to enhance threat detection accuracy in cloud environments. The proposed framework combines Extreme Gradient Boosting (XGBoost) for feature selection with a Convolutional Neural Network–Long Short-Term Memory (CNN-LSTM) architecture for classification. We evaluate the framework using four benchmark datasets: NSL-KDD, CICIDS2017, CSE-CIC-IDS2018, and UNSW-NB15. Experimental results demonstrate that the hybrid approach achieves detection accuracy exceeding 98.5%, with a false positive rate below 1.2% across all datasets. The framework exhibits strong generalization capabilities and maintains real-time detection latency suitable for production cloud deployments. These findings suggest that combining spatial feature extraction with temporal sequence modeling provides a robust foundation for next-generation cloud security systems.