what is the most efficient way to write multiple plots to the local drive?

More than 1.5 million plots must be saved to my local drive before my deep network trained.

What is the best way to write this bunch of plots efficiently?

is there any way to write with graphics (GPU)?

here is my code.

figure('visible','off','position',[0,0,244,244]);
parfor i = 1:numel(data)
        line_width = 1;

        clf,hold on;axis off;
        for j = 1:numLines
            plot((data{i,1}(j,:)),'LineWidth',line_width);
        end
        
        imFileName = [num2str(i) '.jpeg'];
        exportgraphics(gca,imFileName);
end

data is a cell with size of 1000000*1

and each cell size is 10*100

All of them need to be saved not to plotted , but it takes many days!

[dpb]

"More than 1.5 million plots must be saved...."

I fail to understand why the plots would have to be saved and what possible use could be made of that many plots even if did?

What do you really need -- the x,y pairs of data inside the plots, maybe? Even if so, there has to be a much more efficient way to approach the end problem, whatever it is...

[Abolfazl Nejatian]

I am analyzing forex data using deep learning methods.

My goal is to build up network training data.

For this purpose, I have focused on one-minute candles.

At first, I started with analysis in the frequency domain, but unfortunately I did not get a good answer.

After that, I decided to use the Gaussian Regression model to smooth the past of the market in a multi-resolution manner.

the features became more separatebale for the network.

Now I want to teach these features to the network in the form of continuous images.

For this, I am using a hybrid GoogleNet - LSTM.

Answer 1

[Sulaymon Eshkabilov]

In terms of time efficiency, exportgraphics() is faster than saveas().

[Abolfazl Nejatian]

tnx

Answer 2

[Walter Roberson]

Some performance tune-ups

parfor i = 1:numel(data)
    persistent fig ax lines
    line_width = 1;
    if isempty(fig)
        fig = figure('visible','off','position',[0,0,244,244]);
        ax = axes(fig);
    end
    cla(ax);
    plot(ax, data{i}.', 'LineWidth', line_width);  %plot each ROW
    axis(ax, 'off');

    imFileName = i + ".jpeg";
    exportgraphics(ax,imFileName);
end

[Abolfazl Nejatian]

Dear Walter,

Thank you for your comments.

because of the

parfor

this error arised

Error: Persistent variable declarations are not supported in
parfor-loops. For more information, see Parallel for Loops in
    MATLAB, "Global and Persistent Variables".
    
    

And after that I took persistance varibale out of the loop

    persistent fig ax lines
    parfor i = 1:numel(data)
        if isempty(fig)
            fig = figure('visible','off','position',[0,0,244,244]);
            ax = axes(fig);
        end
        cla(ax);
        plot(ax, data{i}.', 'LineWidth', 1);  %plot each ROW
        axis(ax, 'off');
        imFileName = i + ".jpeg";
        exportgraphics(ax,imFileName);
    end
    

and the error was

    The temporary variable 'fig' must be set before it is used. For more
    information, see Parallel for Loops in MATLAB, "Uninitialized
    Temporaries".
    

then i try this situation

persistent fig ax lines
if isempty(fig)
    fig = figure('visible','off','position',[0,0,244,244]);
    ax = axes(fig);
end
parfor i = 1:numel(data)

    cla(ax);
    plot(ax, data{i}.', 'LineWidth', 1);  %plot each ROW
    axis(ax, 'off');
    imFileName = i + ".jpeg";
    exportgraphics(ax,imFileName);
end

and the error is strange for me

Error using exportgraphics
Matrix dimensions must agree.

and the second question is although we have set the size of the figure to 244*244, this size has changed in the output images. how can handle this problem?

My sincere thanks go out to you for your assistance.