Journal of Infrastructure Preservation and Resilience

Table 5 Hyperparameter space for the models developed with Strategy B

From: Modeling retroreflectivity degradation of pavement markings across the US with advanced machine learning algorithms

DT models		ANN models
Hyperparameters	Values	Hyperparameters	Values
Max depth (D)	4, 5, 10, 20, 30, 40, 50	Number of hidden layers (H)	1, 2
Minimum samples split (S)	2, 5, 10, 12, 16, 18, 20, 50, 100	Neurons (N)	15, 30, 60
Minimum samples leaf (L)	2, 5, 10, 20, 30, 40, 50	Batch size (B)	16, 32, 64, 200
Max features (F)	auto, sqrt, log2	Activation function (A)	ReLU, sigmoid, tanh
		Optimizer (O)	adam, SGD
		Learning rate (Lr)	0.001, 0.01, 0.1
		Alpha (α)	0.1, 0.4, 0.6, 0.8

Back to article page