<< ann_getToolboxPath Neural_Network_Functions knn >>

NeuralNet >> Neural_Network_Functions > initialize_parameters

initialize_parameters

Initialize Network Object for Feed-forward Back-Propagation Network (New)

Parameters

N :

Network topology, in 1 x m layers including input and output layers, [Input layer nodes, Hidden layer nodes, Output layer nodes]. If the network is more than one hidden layer, repeat with another number in the "Hidden layer nodes" part.

r :

Range of the weights initial values

af :

Activation functions for each layers

lr :

Learning rate

cost_type :

Function to calculate cost, currently support "mse", "crossentropy", "categorical_crossentropy"

l2 :

L2 regularization

mini_batch :

Minibatch size

Description

Initialize Network Object for Feed-forward Back-Propagation Network Base on Andrew Ng's Coursera Deep-Learning Specialization Course.

Examples

net = initialize_parameters([3 10 2]);

See also

Authors


Report an issue
<< ann_getToolboxPath Neural_Network_Functions knn >>