Enhancing Data Security in Cloud Computing with Optimized Feature Selection and Machine Learning for Intrusion Detection

Abstract

The major objective of this project is to create a Machine Learning (ML) model that can improve data security when data is transported or handled utilizing cloud computing. Researchers must develop a model or a strategy that can secure the data because cloud computing is one of the fastest-growing technologies. This study employs a data set made up of several parameters that can help ML models improve the security of the data for this goal. Then, using a variety of methods, this data set is prepossessed. Following feature extraction, when the parameters are left out, the preprocessed data is then provided to the message learning models. The ML models are then trained and tested using these exclusive parameters. For greater efficiency, three separate learning metrics and three different ML models are created. They are the Artificial Neural Network (ANN) algorithm, the K-nearest neighbor (KNN) algorithm, and the Random Forest (RF) algorithm. The models' testing and training are then used to compare their efficacy. For this reason, the outcomes of both training and testing are noted and examined. The ideal measuring methodology for improving data security is found using parameters like precision, true positive rate, false negative rate, etc. The effectiveness of the feature extraction technique is tabulated alongside the ML models to determine the optimal feature extraction technique and ML model combination. In the end, it is discovered that the ANN algorithm combined with a simulated bee colony may generate the highest level of efficiency. This model's output has a final accuracy of 93.8%. As a result, this approach can be used in conjunction with applications that call for cloud computing, improving data security in the process.