Zero-Day attacks Detection in Cloud using a Hybrid Machine learning Classifier

Cloud computing has become an essential part of the modern IT ecosystem in present time due to its flexibility, openness, accessibility and availability for all. And because of this feature has made it vulnerable to several security threats, including Denial of Service (DoS) and Distributed Denial of Service (DDoS) attacks. Signature and rule-based Intrusion detection system were not able to detect the new and unknown attacks that could be launched against a cloud environment. This research uses machine learning algorithms such as support vector machines (SVM), K-nearest neighbors (KNN), and C4.5 Decision Trees to identify how well specific machine learning algorithms can detect and mitigate security threats. In this study the NSL-KDD Dataset was used. After evaluating the performance of individual classifiers, we ensemble Support Vector Machine, K-nearest neighbors (KNN) and C4.5 Decision Tree. This Hybrid approach has been trained and tested with the above dataset after feature selection SelectKBest and ANOVA F-Test, achieving high accuracy, higher Precision, recall, and F1- score than the individual classifier. The model also shows a low false-positive rate and falsenegative rate. This shows that combining the strength of C4.5 with the classification power of SVM and KNN yields a more reliable and generalizable intrusion detection mechanism.