Skip to main content

Table 4 Hyperparameters of the three models forming the MEE, and their range used by the grid search method

From: A multi-expert ensemble system for predicting Alzheimer transition using clinical features

 

Models

Hyperparameters

Range

Ensemble proposed

Neural Net

Optimizer

{‘SGD’, ‘Adagrad’, ‘Adadelta’, ‘Adam’, ‘Adamax’, ‘Nadam’}

Batch size

{10, 20, 40, 60, 80, 100}

Epochs

{10, 50, 100}

Number of hidden units

{2:2:50}

Random Forest

Max depth

{5, 20, 50, 80, 110}

Min samples for leaf

{3, 4, 5, 10]}

Min samples for split

{8, 10, 12, 24, 32}

Number of estimators

{30, 200, 300, 1000}

Support Vector Machine

C parameter

{0.1,1, 10, 100}

\(\gamma\) parameter

{1, 0.1, 0.01, 0.001}

Kernel

{‘linear’, ‘poly’, ‘rbf’, ‘sigmoid’}