Protecting Cloud Computing from Various Attacks Using Deep Learning Algorithms

Abstract

This study takes a close look on how well deep learning models can find cyber threats in Cloud Based Intrusion Detection Systems. Increasing amounts of organizations are using cloud services, and as this number grows many organizations need an effective and resilient way to identify intrusions when cyber threats continue to grow in number and sophistication. Our research aimed to analyze how good different deep learning models are at detecting and classifying cyber-attacks that occur in a cloud-based environment. In order to achieve this objective, we utilized the NSL-KDD dataset, with two deep learning algorithms (Long Short Term Memory LSTM and Conventional Neural Network). Also, we applied a number of different performance measurements (accuracy, recall, precision, F1-score, confusion matrix), in order to determine how applicable they are to realworld intrusion detection systems. The results showed that, LSTM achieved the highest performance, while still having a solid balance between both accuracy and precision/recall. Our results demonstrate the enormous promise of ensemble and gradient boosting techniques in helping to increase the detection capabilities of cloud based IDSs. Our results illustrate the importance of deep learning in increasing the reliability, robustness, and efficiency of intrusion detection systems used in cloud environments, and ultimately contributes significantly to the security of our digital ecosystems.