Gaussian process regression-based learning rate optimization in convolutional neural networks for medical images classification

Highlights：

• Propose two novel learning rate optimization algorithms for a popular optimizer.

• Apply the Gaussian Process Regression method to predict the classification accuracy.

• Obtain a robust learning rate updating rule during the classification process.

• Study new medical images that map the lipids and proteins without destroying tissue.

• Demonstrate the superiority of the proposed algorithms for classification.