Bayesian Optimization of SVM for Efficient Detection of Cloud Computing Failures

As cloud computing has become a fundamental technology underpinning a wide array of online services, ensuring its stability and reliability has emerged as a critical concern. Fault detection plays a pivotal role in maintaining service continuity and performance. However, the dynamic nature and inherent complexity of cloud computing environments pose significant challenges to effective fault detection. Despite the contributions of existing fault detection methods, they are often hampered by limitations in efficiency and accuracy. This paper proposes an online fault detection model based on Bayesian Optimization (BO) that adjusts the parameters of the Support Vector Machine (SVM) model to improve the efficiency and accuracy of fault detection. Pearson correlation and Principal Component Analysis (PCA) methods are employed to select appropriate attributes from cloud system operational data, thereby optimizing the model and improving detection performance. The experimental results prove that the BO-SVM has achieved an accuracy rate of 93% on the training set. The method we proposed demonstrates higher Accuracy, Precision, Recall, and F-score compared to several representative existing fault detection methods. This paper not only demonstrates the practicality of Bayesian Optimization (BO) in adjusting SVM model hyperparameters but also provides an effective solution for cloud computing fault detection.