An Analysis of Classification Algorithms for Cloud-Based Intrusion Detection Systems

Safeguarding computer networks from unauthorized access and malicious activities heavily relies on Intrusion Detection Systems (IDS), especially in cloud environments where data security is paramount. The effectiveness of an IDS depends heavily on the classification algorithms used to distinguish between legitimate and malicious network traffic. This paper aims to comprehensively study various classification algorithms used in IDS for intrusion detection within cloud environments. It compares the performance of various classification algorithms commonly employed in intrusion detection systems. Random Forest, Logistic Regression, Naive Bayes, Random Tree, support vector machines, and K-Nearest Neighbour are among the algorithms that have been reviewed. The algorithms have also been evaluated using key metrics, including accuracy, error rate, recall, and FValue. The findings provide valuable insights into the strengths and weaknesses of different classification algorithms, aiding in selecting and optimizing cloud-based IDS algorithms for network security. The proposed study employs two feature selection approaches on the NSL-KDD dataset to compare the various classification algorithms for intrusion detection. The results show that the Random Forest classifier with a feature selection of Principal Component Analysis achieves exceptional precision and minimal error, indicating its effectiveness for cloud-based intrusion detection.