How to create Multiplication Layer in deep learning toolbox

Dear community,

I am trying to create multiplication layer which multiplies inputs from multiple neural network layers element-wise, something like "additionLayer to add inputs from multiple layers element-wise". Is there any way I can create this layer in Matlab 2018a?

Best Regards,

Answer 1

[CARLOS VIDAL]

Hi,

I don't think matlab already has available the layer you are looking for, but you can create one.

Follow the link below.

https://www.mathworks.com/help/deeplearning/ug/define-custom-layer-with-multiple-inputs.html

I did it quickly, it seems to work, but better if you double check and follow the instruction in the link above to know how to integrate it with your NN structure.

Hope this will help you.

classdef ElementWiseMultiplication < nnet.layer.Layer

% Example custom ElementWiseMultiplication layer.

properties (Learnable)

% Layer learnable parameters

% Scaling coefficients

end

methods

function layer = ElementWiseMultiplication(numInputs,name)

% layer = ElementWiseMultiplication(numInputs,name) creates a

% element wise multiplication and specifies the number of inputs

% and the layer name.

% Set number of inputs.

layer.NumInputs = numInputs;

% Set layer name.

layer.Name = name;

% Set layer description.

layer.Description = "Element Wise Multiplication of " + numInputs + ...

" inputs";

end

function Z = predict(~, X1,X2)

% Z = predict(layer, X1, ..., Xn) forwards the input data X1,

% ..., Xn through the layer and outputs the result Z.

% Element Wise Multiplication

Z = X1 .*X2;

end

end

end

[ruo li]

Do I need to define a backward function?Thanks.

[Abdelwahab Afifi]

Have you get such layer? If yes, please share the code.