The Design and Experiment of a Diagnostic Algorithm Based on Bayesian Optimization CNN

This study proposes a Bayesian optimization-based diagnostic algorithm for convolutional neural networks (CNN), which improves the predictive performance of the model by optimizing the hyperparameters of the CNN. Traditional CNN models often rely on manual experience to adjust the hyperparameters when facing complex data, which is not only time-consuming but also difficult to ensure finding the optimal solution. To overcome this challenge, this paper adopts the Bayesian optimization algorithm to automatically adjust the hyperparameters of the CNN. Bayesian optimization significantly improves the efficiency of hyperparameter search and the accuracy of the model by constructing an agent model and using the existing experimental data to speculate the next optimal hyperparameter combination. The experimental results show that the CNN model with Bayesian optimization outperforms the standard CNN on the test set, especially in several evaluation indexes such as precision, recall and F1 score, which have achieved significant improvement. In addition, the error estimation in the Bayesian optimization process shows that the estimated value of the minimum error is gradually close to the actual value with the increase of the number of iterations, which verifies the improvement of the accuracy of Bayesian optimization in the optimization process.