Is there any way i generate documentation or comments along with the hdl code using hdl coder?Is there any way i generate documentation or comments along with the hdl code using hdl coder? Is there any way i generate documentation or comments along with the hdl code using hdl coder? hdlcoder, dsp hdl toolbox, hdlverifier MATLAB Answers — New Questions

​

function load_factors_analysis()

% Load the data (assuming ‘load_factors.csv’ is in the working directory)
data = load(‘load_factors.csv’);

% Get central and dispersion measurements for each variable
[mean, median, stddev, min_val, max_val] = central_dispersion_measurements(data);

% Create histograms and/or bar graphs for each variable
histograms_and_bar_graphs(data);

% Create a boxplot (assuming data(:,1) is continuous and data(:,2) is categorical)
boxplot_categorical(data(:, 1), data(:, 2));

% Perform a regression analysis (assuming data(:,1) and data(:,2) are continuous)
regression_analysis(data(:, 1), data(:, 2));

% Discuss the results (replace with your actual analysis)
discussion(data, mean, median, stddev, min_val, max_val);

end

function [mean, median, stddev, min_val, max_val] = central_dispersion_measurements(data)

% Get central and dispersion statistics
mean = mean(data);
median = median(data);
stddev = std(data);
min_val = min(data);
max_val = max(data);

% Return the calculated values
return (mean, median; stddev; min_val; max_val);

end

function histograms_and_bar_graphs(data)

for i = 1:size(data, 2)
if isnumeric(data(:, i))
h = histogram(data(:, i));
title(strcat(‘Histogram of ‘, data(1, i)));
show(h);
else
b = bar(unique(data(:, i)), count(data(:, i)));
title(strcat(‘Bar graph of ‘, data(1, i)));
show(b);
end
end

end

function boxplot_categorical(data_continuous, data_categorical)

% Create a boxplot for continuous data divided by categorical classes
boxplot(data_continuous, data_categorical);

end

function regression_analysis(data1, data2)

% Perform linear regression
[b, a, rsq, pval, se] = regress(data1, data2);

% Print the results
disp(‘Coefficients:’);
disp([b, a]);
disp(‘R-squared:’);
disp(rsq);
disp(‘P-value:’);
disp(pval);
disp(‘Standard error:’);
disp(se);

end

function discussion(~, ~, ~, ~, ~, ~)

% Replace this with your analysis of the data and calculated statistics

disp(‘This is a placeholder for your data analysis discussion.’);
disp(‘Here, you would discuss insights from the central tendency’);
disp(‘(mean, median), dispersion (standard deviation),’);
disp(‘minimum and maximum values, and any relationships found’);
disp(‘between variables using the boxplot and regression analysis.’);

endfunction load_factors_analysis()

% Load the data (assuming ‘load_factors.csv’ is in the working directory)
data = load(‘load_factors.csv’);

% Get central and dispersion measurements for each variable
[mean, median, stddev, min_val, max_val] = central_dispersion_measurements(data);

% Create histograms and/or bar graphs for each variable
histograms_and_bar_graphs(data);

% Create a boxplot (assuming data(:,1) is continuous and data(:,2) is categorical)
boxplot_categorical(data(:, 1), data(:, 2));

% Perform a regression analysis (assuming data(:,1) and data(:,2) are continuous)
regression_analysis(data(:, 1), data(:, 2));

% Discuss the results (replace with your actual analysis)
discussion(data, mean, median, stddev, min_val, max_val);

end

function [mean, median, stddev, min_val, max_val] = central_dispersion_measurements(data)

% Get central and dispersion statistics
mean = mean(data);
median = median(data);
stddev = std(data);
min_val = min(data);
max_val = max(data);

% Return the calculated values
return (mean, median; stddev; min_val; max_val);

end

function histograms_and_bar_graphs(data)

for i = 1:size(data, 2)
if isnumeric(data(:, i))
h = histogram(data(:, i));
title(strcat(‘Histogram of ‘, data(1, i)));
show(h);
else
b = bar(unique(data(:, i)), count(data(:, i)));
title(strcat(‘Bar graph of ‘, data(1, i)));
show(b);
end
end

end

function boxplot_categorical(data_continuous, data_categorical)

% Create a boxplot for continuous data divided by categorical classes
boxplot(data_continuous, data_categorical);

end

function regression_analysis(data1, data2)

% Perform linear regression
[b, a, rsq, pval, se] = regress(data1, data2);

% Print the results
disp(‘Coefficients:’);
disp([b, a]);
disp(‘R-squared:’);
disp(rsq);
disp(‘P-value:’);
disp(pval);
disp(‘Standard error:’);
disp(se);

end

function discussion(~, ~, ~, ~, ~, ~)

% Replace this with your analysis of the data and calculated statistics

disp(‘This is a placeholder for your data analysis discussion.’);
disp(‘Here, you would discuss insights from the central tendency’);
disp(‘(mean, median), dispersion (standard deviation),’);
disp(‘minimum and maximum values, and any relationships found’);
disp(‘between variables using the boxplot and regression analysis.’);

end function load_factors_analysis()

% Load the data (assuming ‘load_factors.csv’ is in the working directory)
data = load(‘load_factors.csv’);

% Get central and dispersion measurements for each variable
[mean, median, stddev, min_val, max_val] = central_dispersion_measurements(data);

% Create histograms and/or bar graphs for each variable
histograms_and_bar_graphs(data);

% Create a boxplot (assuming data(:,1) is continuous and data(:,2) is categorical)
boxplot_categorical(data(:, 1), data(:, 2));

% Perform a regression analysis (assuming data(:,1) and data(:,2) are continuous)
regression_analysis(data(:, 1), data(:, 2));

% Discuss the results (replace with your actual analysis)
discussion(data, mean, median, stddev, min_val, max_val);

end

function [mean, median, stddev, min_val, max_val] = central_dispersion_measurements(data)

% Get central and dispersion statistics
mean = mean(data);
median = median(data);
stddev = std(data);
min_val = min(data);
max_val = max(data);

% Return the calculated values
return (mean, median; stddev; min_val; max_val);

end

function histograms_and_bar_graphs(data)

for i = 1:size(data, 2)
if isnumeric(data(:, i))
h = histogram(data(:, i));
title(strcat(‘Histogram of ‘, data(1, i)));
show(h);
else
b = bar(unique(data(:, i)), count(data(:, i)));
title(strcat(‘Bar graph of ‘, data(1, i)));
show(b);
end
end

end

function boxplot_categorical(data_continuous, data_categorical)

% Create a boxplot for continuous data divided by categorical classes
boxplot(data_continuous, data_categorical);

end

function regression_analysis(data1, data2)

% Perform linear regression
[b, a, rsq, pval, se] = regress(data1, data2);

% Print the results
disp(‘Coefficients:’);
disp([b, a]);
disp(‘R-squared:’);
disp(rsq);
disp(‘P-value:’);
disp(pval);
disp(‘Standard error:’);
disp(se);

end

function discussion(~, ~, ~, ~, ~, ~)

% Replace this with your analysis of the data and calculated statistics

disp(‘This is a placeholder for your data analysis discussion.’);
disp(‘Here, you would discuss insights from the central tendency’);
disp(‘(mean, median), dispersion (standard deviation),’);
disp(‘minimum and maximum values, and any relationships found’);
disp(‘between variables using the boxplot and regression analysis.’);

end statistics, code issues MATLAB Answers — New Questions

​

When click "Send to simulink" Button on Bikesim, i receive and signal like this. How can I solve this problem?
Thanks for your helps.When click "Send to simulink" Button on Bikesim, i receive and signal like this. How can I solve this problem?
Thanks for your helps. When click "Send to simulink" Button on Bikesim, i receive and signal like this. How can I solve this problem?
Thanks for your helps. bikesim, cannot find matlab MATLAB Answers — New Questions

​

Hello MATLAB community,
I’m working on ECG classification using MATLAB, and my data is stored in CSV format with 13 columns (time + 12 leads). I’m interested in extracting PQRST segments using the segmentation method described in this MATLAB code: ECG Segmentation and Filtering.
I have a few questions regarding preprocessing:
Should I apply low-pass filters to all recordings to enhance signal quality?
I intend to retain various types of noise in the data to ensure model robustness. What are your recommendations on managing noise while preserving the integrity of PQRST segments?
Additionally, I’m curious about segmentation approach:
Should I perform segmentation separately for each of the 12 leads, or should I merge them before segmentation? What would be the pros and cons of each approach?
Here an example of an ECG data:

Any insights or suggestions on adapting the segmentation method and handling data preprocessing would be greatly appreciated. Thank you!Hello MATLAB community,
I’m working on ECG classification using MATLAB, and my data is stored in CSV format with 13 columns (time + 12 leads). I’m interested in extracting PQRST segments using the segmentation method described in this MATLAB code: ECG Segmentation and Filtering.
I have a few questions regarding preprocessing:
Should I apply low-pass filters to all recordings to enhance signal quality?
I intend to retain various types of noise in the data to ensure model robustness. What are your recommendations on managing noise while preserving the integrity of PQRST segments?
Additionally, I’m curious about segmentation approach:
Should I perform segmentation separately for each of the 12 leads, or should I merge them before segmentation? What would be the pros and cons of each approach?
Here an example of an ECG data:

Any insights or suggestions on adapting the segmentation method and handling data preprocessing would be greatly appreciated. Thank you! Hello MATLAB community,
I’m working on ECG classification using MATLAB, and my data is stored in CSV format with 13 columns (time + 12 leads). I’m interested in extracting PQRST segments using the segmentation method described in this MATLAB code: ECG Segmentation and Filtering.
I have a few questions regarding preprocessing:
Should I apply low-pass filters to all recordings to enhance signal quality?
I intend to retain various types of noise in the data to ensure model robustness. What are your recommendations on managing noise while preserving the integrity of PQRST segments?
Additionally, I’m curious about segmentation approach:
Should I perform segmentation separately for each of the 12 leads, or should I merge them before segmentation? What would be the pros and cons of each approach?
Here an example of an ECG data:

Any insights or suggestions on adapting the segmentation method and handling data preprocessing would be greatly appreciated. Thank you! ecg, ptb-xl, signal processing, pqrst, segmentation MATLAB Answers — New Questions

​

Hi everyone. First of all, thank you for your time. This will be my first question on the matlab platform. Please excuse me if I have any mistakes. If you understand the problem, you can already find the necessary files in the zip folder. If you want to view images in pgm format, you can use the GIMP application.
I am planning to design a MLP image processing model without using any toolbox.
I plan to train my model by reading one by one 32×30 scale images in the CMU face images dataset I obtained from the internet and then continue with testing process.
(I use imread function that is provided by MATLAB)
INPUT is a cell vector which contains image matrixes in each element. So each element represents an image actually. While processing samples one by one I get its images as column vector.
Here is the code for file operations and image reading:
clc;clear;close;
%*****************Reading Images**************
myFolder = ”; %% Images folder path
if ~isfolder(myFolder) %% Checking if the folder doesn’t exist
errorMessage = sprintf(‘Error: The following folder does not exist:n%snPlease specify a new folder.’, myFolder);
uiwait(warndlg(errorMessage));
myFolder = uigetdir(); % Ask for a new one.
if myFolder == 0
% User clicked Cancel
return;
end
end
filePattern = fullfile(myFolder, ‘*.pgm’);
theFiles = dir(filePattern);
% Define the number of image files in the folder
numImages = length(theFiles);
% Initialize a cell array to store the images
INPUT = cell(numImages, 1);
for k = 1 : length(theFiles)
baseFileName = theFiles(k).name;
fullFileName = fullfile(theFiles(k).folder, baseFileName);
INPUT{k} = imread(fullFileName);
imshow(INPUT{k}); % Display image.
drawnow; % Force display to update immediately.
end
%***********************************************
It is a user-interactive model, and firstly I get the number of:
Hidden layers:
Neurons in each hidden layer: (neuron numbers will be the same for each hidden layer)
Max iteration:
of my model from user.
Other definitions are shown in below code. I define NumberOfInput as 960 which comes from 32×30 because Weight matrix’s size between input layer and first hidden layer needed to adjusted in that way.
Weight matrix values are assigned randomly.
My model should return 0 if person doesn’t wear sunglasses and 1 if person wears with high accuracy. So it is scaler and there will be 1 output obviously.
I studied about MLP models and I found that finding perfect variables is a hard subject in machine learning and it depends on application and tests. So I defined my ETA with various values: 0.01,0.02,0.05,0.2,0.5….
In this type of model input to neurons are defined as netH and output of these neurons are defined as H except last connections. In there they become netO and O.
Also sigma size is defined in order to use after forward state (backward state starts).
My model is an example of Supervised Learning and it needs some outputs for training images like mentioned DESIRED as below. Images inside Model_Training are located as open–>sunglasses–>open–>sunglasses… so I decided to define desired with this order as shown below.
Here is the code:
%*******************VARİABLES*******************
NumberOfPatterns=numImages;
NumberOfInput=960;
NumberOfOutput=1;
LearningRate_ETA=0.5;
while true
NofLayers=input("Layer number: "); % Hidden layer number
Nofneurons=input("Neuron number: "); % Neuron number of each hidden layer
Max_iteration=input("Max iteration number: "); % Max iteration
if(Nofneurons<=0 || NofLayers<=0 || Max_iteration<=0)
fprintf("These values can’t be accepted !");
fprintf("nPlease enter again");
else
break;
end
end
W = cell(NofLayers+1,1);
H=cell(NofLayers,1);
sigma=cell(NofLayers+1,1);
%***********************************************
% Random values are assigned to Weights
for i=1:NofLayers+1
if i==1
W{i}=rand(NumberOfInput,Nofneurons);
elseif i==NofLayers+1
W{i}=rand(Nofneurons,NumberOfOutput);
else
W{i}=rand(Nofneurons,Nofneurons);
end
end
%***********************************************
DESIRED=zeros(NumberOfPatterns,1);
%****************Adjusting Desired Results******
%Training images are located in order. Ex:
%A_open.pgn
%A_sunglasses.pgn
for i=1:NumberOfPatterns
if(mod(i,2)==1)
DESIRED(i)=0;
else
DESIRED(i)=1;
end
end
%************************************************
Right now processing starts. I need to mention that I use sigmoid function as activation function.
In order not to prolong the topic further I will share directly code in here.
%***********************Processing***************
for a=1:Max_iteration
totalerr=0;
for i = 1:NumberOfPatterns
ImageVector = reshape(INPUT{i}, [], 1);
X = double(ImageVector);
for lay=1:NofLayers+1
if(lay==1) %First connections
netH=W{lay}’*X;
H{lay}=sigmoid(netH);%%%
elseif (lay==NofLayers+1) %Last connections
netO=W{lay}’*H{lay-1};
O=sigmoid(netO);
else % between connections layers
netH=W{lay}’*H{lay-1}; %
H{lay}=sigmoid(netH);%
end
end
err=DESIRED(i)-O;
for j=1:NumberOfOutput
sigma{NofLayers+1}=err*O(j)*(1-O(j)); %Last sigma value
end

for l=1:NofLayers
for k=1:Nofneurons
[rowsigma colsigma]=size(sigma{NofLayers-l+2});
[rowW colsW]=size(W{NofLayers-l+2}(k,:));
%These conditions satisfies proper matrix multiplciation
if(colsigma==rowW)
sigma{NofLayers-l+1}=sigma{NofLayers-l+2}*W{NofLayers-l+2}(k,:) *H{NofLayers+1-l}(k)*(1-H{NofLayers+1-l}(k));
else
sigma{NofLayers-l+1}=sigma{NofLayers-l+2}*W{NofLayers-l+2}(k,:)’*H{NofLayers+1-l}(k)*(1-H{NofLayers+1-l}(k));
end
end
end
for z=1:NofLayers+1
%Weights are updated at this part
if((NofLayers+2-z)==1)
W{NofLayers+2-z}=W{NofLayers+2-z}+LearningRate_ETA*X*sigma{NofLayers+2-z};
else
W{NofLayers+2-z}=W{NofLayers+2-z}+LearningRate_ETA*H{NofLayers+1-z}*sigma{NofLayers+2-z};
end
end
totalerr=totalerr+0.5*err^2;
end
cost(a)=totalerr;
end
plot(cost);
%%*****************Test Case********************
%Getting test image address from user
fileFilter = ‘*.pgm’;
[filename, pathname] = uigetfile(fileFilter, ‘Select a PGM file’, ”);
if isequal(filename, 0)
disp(‘Program has stopped’);
else
fullFilePath = fullfile(pathname, filename);
end
%**************Test Sample Operations*******
testSample=imread(fullFilePath); %
testSample=reshape(testSample,[],1);
X=double(testSample);
for lay=1:NofLayers+1
if(lay==1) %First connections
netH=W{lay}’*X;
H{lay}=sigmoid(netH);
elseif (lay==NofLayers+1) %Last connections
netO=W{lay}’*H{lay-1};
Out=round(sigmoid(netO));
else % between connections layers
netH=W{lay}’*H{lay-1};
H{lay}=sigmoid(netH);
end
end
fprintf(‘Result is: %dn’, Out);
%**********************Helper Functions*********
%Sigmoid Activation Function
function y = sigmoid(x)
y = 1 ./ (1 + exp(-x));
end
You can run and test it with the files that provided in zip file. In this kind of model as I know I need to try it with high number of layer and neuron. I tried with 4-20 5-30 5-35 … Generally it returns 1 and this is the problem that I am struggling with.
If you can give any comment, feedback I would appreciate it. Again thank you for giving a time.Hi everyone. First of all, thank you for your time. This will be my first question on the matlab platform. Please excuse me if I have any mistakes. If you understand the problem, you can already find the necessary files in the zip folder. If you want to view images in pgm format, you can use the GIMP application.
I am planning to design a MLP image processing model without using any toolbox.
I plan to train my model by reading one by one 32×30 scale images in the CMU face images dataset I obtained from the internet and then continue with testing process.
(I use imread function that is provided by MATLAB)
INPUT is a cell vector which contains image matrixes in each element. So each element represents an image actually. While processing samples one by one I get its images as column vector.
Here is the code for file operations and image reading:
clc;clear;close;
%*****************Reading Images**************
myFolder = ”; %% Images folder path
if ~isfolder(myFolder) %% Checking if the folder doesn’t exist
errorMessage = sprintf(‘Error: The following folder does not exist:n%snPlease specify a new folder.’, myFolder);
uiwait(warndlg(errorMessage));
myFolder = uigetdir(); % Ask for a new one.
if myFolder == 0
% User clicked Cancel
return;
end
end
filePattern = fullfile(myFolder, ‘*.pgm’);
theFiles = dir(filePattern);
% Define the number of image files in the folder
numImages = length(theFiles);
% Initialize a cell array to store the images
INPUT = cell(numImages, 1);
for k = 1 : length(theFiles)
baseFileName = theFiles(k).name;
fullFileName = fullfile(theFiles(k).folder, baseFileName);
INPUT{k} = imread(fullFileName);
imshow(INPUT{k}); % Display image.
drawnow; % Force display to update immediately.
end
%***********************************************
It is a user-interactive model, and firstly I get the number of:
Hidden layers:
Neurons in each hidden layer: (neuron numbers will be the same for each hidden layer)
Max iteration:
of my model from user.
Other definitions are shown in below code. I define NumberOfInput as 960 which comes from 32×30 because Weight matrix’s size between input layer and first hidden layer needed to adjusted in that way.
Weight matrix values are assigned randomly.
My model should return 0 if person doesn’t wear sunglasses and 1 if person wears with high accuracy. So it is scaler and there will be 1 output obviously.
I studied about MLP models and I found that finding perfect variables is a hard subject in machine learning and it depends on application and tests. So I defined my ETA with various values: 0.01,0.02,0.05,0.2,0.5….
In this type of model input to neurons are defined as netH and output of these neurons are defined as H except last connections. In there they become netO and O.
Also sigma size is defined in order to use after forward state (backward state starts).
My model is an example of Supervised Learning and it needs some outputs for training images like mentioned DESIRED as below. Images inside Model_Training are located as open–>sunglasses–>open–>sunglasses… so I decided to define desired with this order as shown below.
Here is the code:
%*******************VARİABLES*******************
NumberOfPatterns=numImages;
NumberOfInput=960;
NumberOfOutput=1;
LearningRate_ETA=0.5;
while true
NofLayers=input("Layer number: "); % Hidden layer number
Nofneurons=input("Neuron number: "); % Neuron number of each hidden layer
Max_iteration=input("Max iteration number: "); % Max iteration
if(Nofneurons<=0 || NofLayers<=0 || Max_iteration<=0)
fprintf("These values can’t be accepted !");
fprintf("nPlease enter again");
else
break;
end
end
W = cell(NofLayers+1,1);
H=cell(NofLayers,1);
sigma=cell(NofLayers+1,1);
%***********************************************
% Random values are assigned to Weights
for i=1:NofLayers+1
if i==1
W{i}=rand(NumberOfInput,Nofneurons);
elseif i==NofLayers+1
W{i}=rand(Nofneurons,NumberOfOutput);
else
W{i}=rand(Nofneurons,Nofneurons);
end
end
%***********************************************
DESIRED=zeros(NumberOfPatterns,1);
%****************Adjusting Desired Results******
%Training images are located in order. Ex:
%A_open.pgn
%A_sunglasses.pgn
for i=1:NumberOfPatterns
if(mod(i,2)==1)
DESIRED(i)=0;
else
DESIRED(i)=1;
end
end
%************************************************
Right now processing starts. I need to mention that I use sigmoid function as activation function.
In order not to prolong the topic further I will share directly code in here.
%***********************Processing***************
for a=1:Max_iteration
totalerr=0;
for i = 1:NumberOfPatterns
ImageVector = reshape(INPUT{i}, [], 1);
X = double(ImageVector);
for lay=1:NofLayers+1
if(lay==1) %First connections
netH=W{lay}’*X;
H{lay}=sigmoid(netH);%%%
elseif (lay==NofLayers+1) %Last connections
netO=W{lay}’*H{lay-1};
O=sigmoid(netO);
else % between connections layers
netH=W{lay}’*H{lay-1}; %
H{lay}=sigmoid(netH);%
end
end
err=DESIRED(i)-O;
for j=1:NumberOfOutput
sigma{NofLayers+1}=err*O(j)*(1-O(j)); %Last sigma value
end

for l=1:NofLayers
for k=1:Nofneurons
[rowsigma colsigma]=size(sigma{NofLayers-l+2});
[rowW colsW]=size(W{NofLayers-l+2}(k,:));
%These conditions satisfies proper matrix multiplciation
if(colsigma==rowW)
sigma{NofLayers-l+1}=sigma{NofLayers-l+2}*W{NofLayers-l+2}(k,:) *H{NofLayers+1-l}(k)*(1-H{NofLayers+1-l}(k));
else
sigma{NofLayers-l+1}=sigma{NofLayers-l+2}*W{NofLayers-l+2}(k,:)’*H{NofLayers+1-l}(k)*(1-H{NofLayers+1-l}(k));
end
end
end
for z=1:NofLayers+1
%Weights are updated at this part
if((NofLayers+2-z)==1)
W{NofLayers+2-z}=W{NofLayers+2-z}+LearningRate_ETA*X*sigma{NofLayers+2-z};
else
W{NofLayers+2-z}=W{NofLayers+2-z}+LearningRate_ETA*H{NofLayers+1-z}*sigma{NofLayers+2-z};
end
end
totalerr=totalerr+0.5*err^2;
end
cost(a)=totalerr;
end
plot(cost);
%%*****************Test Case********************
%Getting test image address from user
fileFilter = ‘*.pgm’;
[filename, pathname] = uigetfile(fileFilter, ‘Select a PGM file’, ”);
if isequal(filename, 0)
disp(‘Program has stopped’);
else
fullFilePath = fullfile(pathname, filename);
end
%**************Test Sample Operations*******
testSample=imread(fullFilePath); %
testSample=reshape(testSample,[],1);
X=double(testSample);
for lay=1:NofLayers+1
if(lay==1) %First connections
netH=W{lay}’*X;
H{lay}=sigmoid(netH);
elseif (lay==NofLayers+1) %Last connections
netO=W{lay}’*H{lay-1};
Out=round(sigmoid(netO));
else % between connections layers
netH=W{lay}’*H{lay-1};
H{lay}=sigmoid(netH);
end
end
fprintf(‘Result is: %dn’, Out);
%**********************Helper Functions*********
%Sigmoid Activation Function
function y = sigmoid(x)
y = 1 ./ (1 + exp(-x));
end
You can run and test it with the files that provided in zip file. In this kind of model as I know I need to try it with high number of layer and neuron. I tried with 4-20 5-30 5-35 … Generally it returns 1 and this is the problem that I am struggling with.
If you can give any comment, feedback I would appreciate it. Again thank you for giving a time. Hi everyone. First of all, thank you for your time. This will be my first question on the matlab platform. Please excuse me if I have any mistakes. If you understand the problem, you can already find the necessary files in the zip folder. If you want to view images in pgm format, you can use the GIMP application.
I am planning to design a MLP image processing model without using any toolbox.
I plan to train my model by reading one by one 32×30 scale images in the CMU face images dataset I obtained from the internet and then continue with testing process.
(I use imread function that is provided by MATLAB)
INPUT is a cell vector which contains image matrixes in each element. So each element represents an image actually. While processing samples one by one I get its images as column vector.
Here is the code for file operations and image reading:
clc;clear;close;
%*****************Reading Images**************
myFolder = ”; %% Images folder path
if ~isfolder(myFolder) %% Checking if the folder doesn’t exist
errorMessage = sprintf(‘Error: The following folder does not exist:n%snPlease specify a new folder.’, myFolder);
uiwait(warndlg(errorMessage));
myFolder = uigetdir(); % Ask for a new one.
if myFolder == 0
% User clicked Cancel
return;
end
end
filePattern = fullfile(myFolder, ‘*.pgm’);
theFiles = dir(filePattern);
% Define the number of image files in the folder
numImages = length(theFiles);
% Initialize a cell array to store the images
INPUT = cell(numImages, 1);
for k = 1 : length(theFiles)
baseFileName = theFiles(k).name;
fullFileName = fullfile(theFiles(k).folder, baseFileName);
INPUT{k} = imread(fullFileName);
imshow(INPUT{k}); % Display image.
drawnow; % Force display to update immediately.
end
%***********************************************
It is a user-interactive model, and firstly I get the number of:
Hidden layers:
Neurons in each hidden layer: (neuron numbers will be the same for each hidden layer)
Max iteration:
of my model from user.
Other definitions are shown in below code. I define NumberOfInput as 960 which comes from 32×30 because Weight matrix’s size between input layer and first hidden layer needed to adjusted in that way.
Weight matrix values are assigned randomly.
My model should return 0 if person doesn’t wear sunglasses and 1 if person wears with high accuracy. So it is scaler and there will be 1 output obviously.
I studied about MLP models and I found that finding perfect variables is a hard subject in machine learning and it depends on application and tests. So I defined my ETA with various values: 0.01,0.02,0.05,0.2,0.5….
In this type of model input to neurons are defined as netH and output of these neurons are defined as H except last connections. In there they become netO and O.
Also sigma size is defined in order to use after forward state (backward state starts).
My model is an example of Supervised Learning and it needs some outputs for training images like mentioned DESIRED as below. Images inside Model_Training are located as open–>sunglasses–>open–>sunglasses… so I decided to define desired with this order as shown below.
Here is the code:
%*******************VARİABLES*******************
NumberOfPatterns=numImages;
NumberOfInput=960;
NumberOfOutput=1;
LearningRate_ETA=0.5;
while true
NofLayers=input("Layer number: "); % Hidden layer number
Nofneurons=input("Neuron number: "); % Neuron number of each hidden layer
Max_iteration=input("Max iteration number: "); % Max iteration
if(Nofneurons<=0 || NofLayers<=0 || Max_iteration<=0)
fprintf("These values can’t be accepted !");
fprintf("nPlease enter again");
else
break;
end
end
W = cell(NofLayers+1,1);
H=cell(NofLayers,1);
sigma=cell(NofLayers+1,1);
%***********************************************
% Random values are assigned to Weights
for i=1:NofLayers+1
if i==1
W{i}=rand(NumberOfInput,Nofneurons);
elseif i==NofLayers+1
W{i}=rand(Nofneurons,NumberOfOutput);
else
W{i}=rand(Nofneurons,Nofneurons);
end
end
%***********************************************
DESIRED=zeros(NumberOfPatterns,1);
%****************Adjusting Desired Results******
%Training images are located in order. Ex:
%A_open.pgn
%A_sunglasses.pgn
for i=1:NumberOfPatterns
if(mod(i,2)==1)
DESIRED(i)=0;
else
DESIRED(i)=1;
end
end
%************************************************
Right now processing starts. I need to mention that I use sigmoid function as activation function.
In order not to prolong the topic further I will share directly code in here.
%***********************Processing***************
for a=1:Max_iteration
totalerr=0;
for i = 1:NumberOfPatterns
ImageVector = reshape(INPUT{i}, [], 1);
X = double(ImageVector);
for lay=1:NofLayers+1
if(lay==1) %First connections
netH=W{lay}’*X;
H{lay}=sigmoid(netH);%%%
elseif (lay==NofLayers+1) %Last connections
netO=W{lay}’*H{lay-1};
O=sigmoid(netO);
else % between connections layers
netH=W{lay}’*H{lay-1}; %
H{lay}=sigmoid(netH);%
end
end
err=DESIRED(i)-O;
for j=1:NumberOfOutput
sigma{NofLayers+1}=err*O(j)*(1-O(j)); %Last sigma value
end

for l=1:NofLayers
for k=1:Nofneurons
[rowsigma colsigma]=size(sigma{NofLayers-l+2});
[rowW colsW]=size(W{NofLayers-l+2}(k,:));
%These conditions satisfies proper matrix multiplciation
if(colsigma==rowW)
sigma{NofLayers-l+1}=sigma{NofLayers-l+2}*W{NofLayers-l+2}(k,:) *H{NofLayers+1-l}(k)*(1-H{NofLayers+1-l}(k));
else
sigma{NofLayers-l+1}=sigma{NofLayers-l+2}*W{NofLayers-l+2}(k,:)’*H{NofLayers+1-l}(k)*(1-H{NofLayers+1-l}(k));
end
end
end
for z=1:NofLayers+1
%Weights are updated at this part
if((NofLayers+2-z)==1)
W{NofLayers+2-z}=W{NofLayers+2-z}+LearningRate_ETA*X*sigma{NofLayers+2-z};
else
W{NofLayers+2-z}=W{NofLayers+2-z}+LearningRate_ETA*H{NofLayers+1-z}*sigma{NofLayers+2-z};
end
end
totalerr=totalerr+0.5*err^2;
end
cost(a)=totalerr;
end
plot(cost);
%%*****************Test Case********************
%Getting test image address from user
fileFilter = ‘*.pgm’;
[filename, pathname] = uigetfile(fileFilter, ‘Select a PGM file’, ”);
if isequal(filename, 0)
disp(‘Program has stopped’);
else
fullFilePath = fullfile(pathname, filename);
end
%**************Test Sample Operations*******
testSample=imread(fullFilePath); %
testSample=reshape(testSample,[],1);
X=double(testSample);
for lay=1:NofLayers+1
if(lay==1) %First connections
netH=W{lay}’*X;
H{lay}=sigmoid(netH);
elseif (lay==NofLayers+1) %Last connections
netO=W{lay}’*H{lay-1};
Out=round(sigmoid(netO));
else % between connections layers
netH=W{lay}’*H{lay-1};
H{lay}=sigmoid(netH);
end
end
fprintf(‘Result is: %dn’, Out);
%**********************Helper Functions*********
%Sigmoid Activation Function
function y = sigmoid(x)
y = 1 ./ (1 + exp(-x));
end
You can run and test it with the files that provided in zip file. In this kind of model as I know I need to try it with high number of layer and neuron. I tried with 4-20 5-30 5-35 … Generally it returns 1 and this is the problem that I am struggling with.
If you can give any comment, feedback I would appreciate it. Again thank you for giving a time. mlp, image processing, deep learning MATLAB Answers — New Questions

​

hello, when i try use function caller to call the subfunction, following eror happens, how to resolve it.hello, when i try use function caller to call the subfunction, following eror happens, how to resolve it. hello, when i try use function caller to call the subfunction, following eror happens, how to resolve it. function caller, simulink MATLAB Answers — New Questions

​

I have many vhdl code files. Is there any way i can generate a call graph from MATLAB itself?I have many vhdl code files. Is there any way i can generate a call graph from MATLAB itself? I have many vhdl code files. Is there any way i can generate a call graph from MATLAB itself? hdlcoder, vhdl, dsp MATLAB Answers — New Questions

​