plotwb

(To be removed) Plot Hinton diagram of weight and bias values

plotwb 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

plotwb(net)
plotwb(IW,LW,B)
plotwb(...,'toLayers',toLayers)
plotwb(...,'fromInputs',fromInputs)
plotwb(...,'fromLayers',fromLayers)
plotwb(...,'root',root)

Description

plotwb(net) takes a neural network and plots all its weights and biases.

plotwb(IW,LW,B) takes a neural networks input weights, layer weights and biases and plots them.

plotwb(...,'toLayers',toLayers) optionally defines which destination layers whose input weights, layer weights and biases will be plotted.

plotwb(...,'fromInputs',fromInputs) optionally defines which inputs will have their weights plotted.

plotwb(...,'fromLayers',fromLayers) optionally defines which layers will have weights coming from them plotted.

plotwb(...,'root',root) optionally defines the root used to scale the weight/bias patch sizes. The default is 2, which makes the 2-dimensional patch sizes scale directly with absolute weight and bias sizes. Larger values of root magnify the relative patch sizes of smaller weights and biases, making differences in smaller values easier to see.

Examples

Plot Weights and Biases

Here a cascade-forward network is configured for particular data and its weights and biases are plotted in several ways.

[x,t] = simplefit_dataset;
net = cascadeforwardnet([15 5]);
net = configure(net,x,t);
plotwb(net)

MATLAB figure

plotwb(net,'root',3)

MATLAB figure

plotwb(net,'root',4)

MATLAB figure

plotwb(net,'toLayers',2)

MATLAB figure

plotwb(net,'fromLayers',1)

MATLAB figure

plotwb(net,'toLayers',2,'fromInputs',1)

MATLAB figure

Version History

Introduced in R2010b

collapse all

See Also

| (Statistics and Machine Learning Toolbox) | (Statistics and Machine Learning Toolbox) | | |

Topics