Cloud-based multiclass anomaly detection and categorization using ensemble learning
Faisal Shahzad, Abdul Mannan, Abdul Rehman Javed, Ahmad Almadhor, Thar Baker, Dhiya Al‐Jumeily
Abstract
Abstract The world of the Internet and networking is exposed to many cyber-attacks and threats. Over the years, machine learning models have progressed to be integrated into many scenarios to detect anomalies accurately. This paper proposes a novel approach named cloud-based anomaly detection ( CAD ) to detect cloud-based anomalies. CAD consist of two key blocks: ensemble machine learning (EML) model for binary anomaly classification and convolutional neural network long short-term memory ( CNN-LSTM ) for multiclass anomaly categorization. CAD is evaluated on a complex UNSW dataset to analyze the performance of binary anomaly detection and categorization of multiclass anomalies. Furthermore, the comparison of CAD with other machine learning conventional models and state-of-the-art studies have been presented. Experimental analysis shows that CAD outperforms other studies by achieving the highest accuracy of 97.06% for binary anomaly detection and 99.91% for multiclass anomaly detection.