Train a GAN example difficulty

I am working through the example for using a GAN given here:

https://www.mathworks.com/help/deeplearning/ug/train-generative-adversarial-network.html

And I get an error at the point where it says projectAndReshapeLayer.

Undefined function 'projectAndReshapeLayer' for input arguments of type
'double'.

When I click on the word projectAndReshapeLayer, I get this:

You clicked a link that corresponds to this MATLAB command:

edit(fullfile(matlabroot,'examples','nnet','main','projectAndReshapeLayer.m'))

Run the command by entering it in the MATLAB Command Window. Web browsers do not support MATLAB commands.

So when I paste this link into the Command Window, it simply creates a blank projectAndReshapeLayer.m file and the error persists.

What am I doing wrong?

To project and reshape the noise input, use the custom layer projectAndReshapeLayer, attached to this example as a supporting file. The projectAndReshapeLayer layer upscales the input using a fully connected operation and reshapes the output to the specified size.

filterSize = 5;
numFilters = 64;
numLatentInputs = 100;

projectionSize = [4 4 512];

layersGenerator = [
    imageInputLayer([1 1 numLatentInputs],'Normalization','none','Name','in')
    projectAndReshapeLayer(projectionSize,numLatentInputs,'proj');
    transposedConv2dLayer(filterSize,4*numFilters,'Name','tconv1')
    batchNormalizationLayer('Name','bnorm1')
    reluLayer('Name','relu1')
    transposedConv2dLayer(filterSize,2*numFilters,'Stride',2,'Cropping','same','Name','tconv2')
    batchNormalizationLayer('Name','bnorm2')
    reluLayer('Name','relu2')
    transposedConv2dLayer(filterSize,numFilters,'Stride',2,'Cropping','same','Name','tconv3')
    batchNormalizationLayer('Name','bnorm3')
    reluLayer('Name','relu3')
    transposedConv2dLayer(filterSize,3,'Stride',2,'Cropping','same','Name','tconv4')
    tanhLayer('Name','tanh')];

lgraphGenerator = layerGraph(layersGenerator);

[Dong-Ho]

I have also same problem with you !

Anybody tackle it please.

[Hiroyuki Hishida]

Hi,

Do you stll have this problem? Iy you have, please tell your MATLAB version.

version 

I checked this issue under "9.8.0.1380330 (R2020a) Update 2" but can see "projectAndReshapeLayer" correctly.

HTH

Hiroyuki

[mark palmer]

Im on 2019b, thanks for the explanation.

[SeungRyeol LEE]

Use the livescripts not original scripts

[Alexander Hagg]

I found that file in the Matlab installation folder

R2020b/examples/nnet/main/projectAndReshapeLayer.m

[Yonglei GUI]

classdef projectAndReshapeLayer < nnet.layer.Layer
    properties
        % (Optional) Layer properties.
        OutputSize
    end
    properties (Learnable)
        % Layer learnable parameters.
        
        Weights
        Bias
    end
    
    methods
        function layer = projectAndReshapeLayer(outputSize, numChannels, name)
            % Create a projectAndReshapeLayer.
            
            % Set layer name.
            layer.Name = name;
            % Set layer description.
            layer.Description = "Project and reshape layer with output size " + join(string(outputSize));
            
            % Set layer type.
            layer.Type = "Project and Reshape";
            
            % Set output size.
            layer.OutputSize = outputSize;
            
            % Initialize fully connect weights and bias.
            fcSize = prod(outputSize);
            layer.Weights = initializeGlorot(fcSize, numChannels);
            layer.Bias = zeros(fcSize, 1, 'single');
        end
        
        function Z = predict(layer, X)
            % Forward input data through the layer at prediction time and
            % output the result.
            %
            % Inputs:
            %         layer - Layer to forward propagate through
            %         X     - Input data, specified as a 1-by-1-by-C-by-N 
            %                 dlarray, where N is the mini-batch size.
            % Outputs:
            %         Z     - Output of layer forward function returned as 
            %                 an sz(1)-by-sz(2)-by-sz(3)-by-N dlarray,
            %                 where sz is the layer output size and N is
            %                 the mini-batch size.
            % Fully connect.
            weights = layer.Weights;
            bias = layer.Bias;
            X = fullyconnect(X,weights,bias,'DataFormat','SSCB');
            
            % Reshape.
            outputSize = layer.OutputSize;
            Z = reshape(X, outputSize(1), outputSize(2), outputSize(3), []);
        end
    end
end
function weights = initializeGlorot(numOut, numIn)
% Initialize weights using uniform Glorot.
varWeights = sqrt( 6 / (numIn + numOut) );
weights = varWeights * (2 * rand([numOut, numIn], 'single') - 1);
end

[Rahul Gowtham Poola]

Error using dlnetwork/validateForwardInputs

Layer 'in': Invalid input data. Invalid number of spatial dimensions. Layer expects 2 but received 0.

[zahoor m]

Layer 'input': Invalid input data. Invalid number of spatial dimensions. Layer expects 2 but received

0.

Answer 1

[Xiangxue Wang]

Trying to add the projectAndReshapeLayer path to your matlab searching path. By default, the deep learning example are not in 2020 path.

Hope this will work for you

>> fullfile(matlabroot,'examples','nnet','main','projectAndReshapeLayer.m')

ans =

    'C:\Program Files\MATLAB\R2020a\examples\nnet\main\projectAndReshapeLayer.m'
    
% so adding path to by:
addpath('C:\Program Files\MATLAB\R2020a\examples\nnet\main')

Answer 2

[soliya]

https://viewer.mathworks.com/?viewer=plain_code&url=https%3A%2F%2Fww2.mathworks.cn%2Fmatlabcentral%2Fmlc-downloads%2Fdownloads%2F04c02184-6b40-49d0-be14-8fae65da6ea0%2Fed1525d6-681b-446a-8b35-4501180bc312%2Ffiles%2Ffiles%2FprojectAndReshapeLayer.m&embed=web

[Yonglei GUI]

Hello, my matlab has no the file minibatchqueue.m, so may you give me a URL like this? or you can send email [e-mail]. Thank you very much!

Answer 3

[Ieuan Evans]

This example was updated in R2020a to use this custom layer. If you use the command openExample('nnet/TrainGenerativeAdversarialNetworkGANExample') in MATLAB, then it will open the correct version of this example for your version of MATLAB.

Hope this helps.

Answer 4

[Ramyakrishna]

Replace the line with the below line

projectAndReshapeLayer(projectionSize,numLatentInputs,Name='proj')