ENHANCING CLOUD WORKLOAD SECURITY WITH PREDICTIVE ANALYTICS AND MACHINE LEARNING

Abstract

Improving cloud security requires a significant contribution from machine learning (ML). Organisations may strengthen their cloud infrastructure by proactively detecting, mitigating, and responding to growing cyber threats by using the capabilities of AI and ML. Security systems can now identify trends, abnormalities, and possible risks in large datasets thanks to AI-driven methodologies. By using past attack data, machine learning algorithms may anticipate new attacks and create defences that are more potent. Furthermore, identity management is strengthened by AI-enhanced authentication and access control systems, which lower the possibility of unauthorised access and data breaches. The use of a hybrid prediction strategy based on machine learning methods is the main topic of this study. The goal of the first step is to divide the input signal of the time series data into two halves. Second, Support Vector Regression (SVR) is applied to the first section in order to forecast low frequency components. An artificial neural network (ANN) is employed for prediction since the second section of the time series is more likely to include noise and has a high frequency. Finally, to accomplish accurate workload prediction, an inverse wavelet transformation is used to reconstruct these samples to the original signal from two multi-scale forecasts. According to the overall findings, the suggested strategy is comparatively more predictable than the competing strategy. Two models are used to analyse the outcomes, and the hybrid SVR + ANN model that was suggested performed better than the other model.