revert

(To be removed) Change network weights and biases to previous initialization values

revert will be removed in a future release. For more information, see Transition Legacy Neural Network Code to dlnetwork Workflows.

For advice on updating your code, see Version History.

Syntax

net = revert(net)


Description

net = revert(net) returns neural network net with weight and bias values restored to the values generated the last time the network was initialized.

If the network is altered so that it has different weight and bias connections or different input or layer sizes, then revert cannot set the weights and biases to their previous values and they are set to zeros instead.

Examples

Here a perceptron is created with input size set to 2 and number of neurons to 1.

net = perceptron;
net.inputs{1}.size = 2;
net.layers{1}.size = 1;

The initial network has weights and biases with zero values.

net.iw{1,1}, net.b{1}

Change these values as follows:

net.iw{1,1} = [1 2]; 
net.b{1} = 5;
net.iw{1,1}, net.b{1}

You can recover the network’s initial values as follows:

net = revert(net);
net.iw{1,1}, net.b{1}

Version History

Introduced before R2006a

R2026a: revert will be removed

revert will be removed in a future release.

To learn more about initializing and customizing neural network learnable parameters, see these examples and topics:

• Compare Layer Weight Initializers

• Specify Custom Weight Initialization Function

• Neural Network Weight Tying

• Initialize Learnable Parameters for Model Function

For more information, see Transition Legacy Neural Network Code to dlnetwork Workflows.