Machine Learning-Enhanced IDS: RFE-LSTM-Based Model for Cloud Security
Karthik Rajashekaran, Rafaqut Kazmi, Rahul Jain · International Journal of Computer Trends and Technology · 2024
Cloud computing impacts huge information science due to its services as infrastructure, software services, and platforms. The widespread use of cloud computing presents challenges such as security, privacy, and trust. The main threats are the susceptibility of the cloud infrastructure to various attacks, including address resolution protocol, IP spoofing, and denial of service. The classical intrusion detection techniques are insufficient to mitigate these new threats. The research proposes the REF-LSTM-IDS model, a novel technique that combines Recursive Feature Elimination (RFE) for optimised feature selection with a Long Short-Term Memory (LSTM) network used to identify dynamic threat pattern recognition. The proposed model's performance was assessed on the NSL-KDD and BoT-IoT datasets for feature selection reduction capability, and it was found that the model performs reasonably well on the evaluation criteria of accuracy and precision. The model performed 91.50% and 92.21% for accuracy measures for the datasets provided. The precision measure performance was 47.54%, and the recall measure was 82.31% for the datasets provided for the Matthews Correlation Coefficient (MCC) across the whole dataset. The proposed model improves cloud security and provides new insights for the integrated IDS model with machine learning capabilities. The integrated models reduce emerging security threats with embedded intelligence.