How to simulate trained networks?

Hi guys, I am trying to feed my trained networks with some data. I want to make a plot from its output

clear;
clc;
load network1

layers_encode(1)=network1.Layers(1);
layers_encode(2)=network1.Layers(2);
layers_encode(3)=network1.Layers(3);
layers_encode(4)=network1.Layers(4);
layers_encode(5)=network1.Layers(5);
layers_encode(6)=regressionLayer;

net_encode=assembleNetwork(layers_encode);

[xTrain, tTrain, xValid, tValid, xTest, tTest] = LoadMNIST(3);
xTrain= mat2gray(reshape(xTrain, [28*28 50000]));
inputs=xTrain(:,1:1000);
nInput=size(inputs,2);

[Y]=sim(net_encode, inputs);

scatterplot(Y) %i'd like to compare the input and the output with scatterplot
scatterplot(inputs)

but I cannot get the output.

i tried using [Y]=sim(net,x) but matlab gives 'model' parameter must be a character vector

I read this https://www.mathworks.com/help/deeplearning/ref/sim.html, and tried

y=net_code(inputs), matlab gives Array indices must be positive integers or logical values.

anybody knows what should I do?

Answer 1

[Shraddha Jain]

Hi Yanuar,

I understand that you are facing issues while performing regression on input images using the method,

Y = sim(net_encode, inputs)

However, I am not able to reproduce the error at my end. For in-depth understanding of the issue, could you share the network1 or the details of it.

Meanwhile, you may try using the predict method available in Deep Learning Toolbox as,

Y = predict(net_encode, inputs)

This method predicts responses using a trained deep learning neural network on image data as well. Please refer to this documentation for more information on predict method.

[Yanuar Rizki Pahlevi]

in the end I did used predict and got the answer. thanks for looking at this question!