Table 5 Superparameters and optimized parameters of DNN model.

From: Prediction of the displacements of the pile tops and ground surface around piles based on machine learning algorithms

Type

Learning rate

Regularization coefficient

Activation function

Number of hidden layers

Number of nodes in each hidden layer

Hyperparameters

0.02、0.03、0.04、0.05、0.06、0.07、0.8

0.02、0.03、0.04、0.05、0.06、0.07、0.8

ReLU、Sigmoid、tanh

3、4、5、6

16、32、64、128

Optimized parameters

0.06

0.05

ReLU

4

64

Back to article page