Diagnosis of Retinal Diseases Based on Bayesian Optimization Deep Learning Network Using Optical Coherence Tomography Images

Table 2

Hyperparameters and their search space.

Parameter	Search space	Description
Optimizer	Adam, RMSProp, SGD, AdaDelta	To optimize the input weights by comparing the prediction and the loss function
Learning rate	1e − 3, 1e − 4, 1e − 5, 1e − 6	To determine the step size at each iteration while minimizing the loss function
Activation function	ReLu, Elu and Tanh, Leaky ReLu	To introduce nonlinearity into the output of neurons
Number of neurons in customized layers	64,128, 56, 512,1024	To compute the weighted average of the input
Batch size	32,64,128	Number of training examples utilized in one iteration