svmtrain function is not working in my matlab 2016a.

I want to run svmtrain function on 2016a version of matlab. what will do

Answer 1

[Steven Lord]

Do you have Statistics and Machine Learning Toolbox installed? You can check this using the ver function. Since svmtrain is part of this toolbox, you must have that toolbox installed in order to use this function.

But if you have Statistics and Machine Learning Toolbox included in your installation, you might want to use the newer functions / classes introduced in release R2014a like fitcsvm, ClassificationSVM, and CompactClassificationSVM. This will be especially important if you want your code to work in release R2018a and later, as the Release Notes to which I linked indicate that svmtrain will error starting in release R2018a.

[Sajja Tulasi Krishna]

yes, I installed Statistics and Machine Learning Toolbox. Eventhough matlab 2016a not worked with svmtrain function.

Is svmtrain function is work with 2014a. If not which vesion is supported svm function

[Steven Lord]

Show a small section of code that contains your attempt to call svmtrain in release R2016a. In addition, include the exact output MATLAB displayed (including the full text of any error or warning message you received, all the text in red or orange) when you ran that code.

[Sajja Tulasi Krishna]

%

AUTHOR: Jaime Ivan Cervantes
%LAST UPDATED: December 13, 2011
%DESCRIPTION: This code implements PCA, Fisher LDA, and SVM to find matching faces.

% Clear memory and console
close all
clear
clc

% Choose simple, pose variation, or illumination variation dataset 
dataset = input('Please select your dataset: \n(1) simple.m \n(2) pose.m\n');

if dataset == 1
     load 'E:/Ph.D Work/Softwares/PhD_tool/demos/svm/simple.mat'
     %Define variables
     k = 200;% Number of classes
     n = 3;% Number of images per class 
elseif dataset == 2
     load 'E:/Ph.D Work/Softwares/PhD_tool/demos/svm/pose.mat'
     face = pose;
    %Define variables
     k = 68;% Number of classes
     n = 13;% Number of images per class
else
    display('Input not valid');
end

testNum = input('Enter the number of the face for which you want to find the match: ');
classType = input('Choose (1) for SVM or (2) for Euclidean Distance: ');

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Principal Component Analysis (PCA)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

display('Loading please wait...')

% Read images in T matrix
[nRow nCol M] = size(face);

% T is a matrix containing the reshaped vectors for each image
T = reshape(face,[nRow*nCol M]);

% mTot is the mean of the entire set of training images
mTot = mean(T,2);

% substract mean
A = T-repmat(mTot,1,M);

% Obtaining eigenvalues and eigenvectors of A'A
[V,D] = eig(A'*A);

% Obtaning more relevant eigenvalues and eigenvectors
eval = diag(D);

peval = [];
pevec = [];

for i = M:-1:k+1
    peval = [peval eval(i)];
    pevec = [pevec V(:,i)];
end

% Obtaining the eigenvectors
U = A * pevec; 

% Obtaining PCA weights
 Wpca = U'*A;

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Fisher's Linear Discriminant Analysis
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

% Obtaining Sb and Sw
cMean = zeros(M-k,M-k);
Sb = zeros(M-k,M-k);
Sw = zeros(M-k,M-k);

pcaMean = mean(Wpca,2);

for i = 1:k
    cMean = mean(Wpca(:,n*i-(n-1):n*i),2);
    Sb = Sb + (cMean-pcaMean)*(cMean-pcaMean)';
end

Sb = n*Sb;

for i = 1:k
    cMean = mean(Wpca(:,n*i-(n-1):n*i),2);
    for j = n*i-(n-1):n*i
         Sw = Sw + (Wpca(:,j)-cMean)*(Wpca(:,j)-cMean)';
    end
end

% Obtaining Fisher eigenvectors and eigenvalues
[Vf, Df] = eig(Sb,Sw);

% Calculating weights
 Df = fliplr(diag(Df));
 Vf = fliplr(Vf);

% Calculating fisher weights
Wf = Vf'*Wpca;

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Support Vector Machine (LIBSCM Dependency)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

if classType == 1

      % testNum = 13*20;

      % Reshape the selected face
      Tr = reshape(face(:,:,testNum),[nRow*nCol 1]);
      Ar = Tr-mTot;

      % Obtain the weights of the normalized selected face
      Wrec = Vf'*U'*Ar;

      % SVM parameters. The kernel is a polynomial
      c=1e9;
      params=[' -t ' int2str(1) ' -c ' int2str(c)];

      % prevArray starts with an array containing each class. Winner array is the
      %classes that are selected in the binary tree
      prevArr = [1:k]; 
      winnerArr = [];

      % This section of the code computes a binary SVM tree, by solving a 2 class
      % problem with SVM for every 2 classes (1 and 2, 3 and 4, 5 and 6 etc). The
      % that you selected is classified according to each 2-class problem and the
      % class selected goes on to compete with the other classes selected.
      for max = 1:1000
          winnerArr = []; 
          for winRep = 1:2:length(prevArr) 

              % Selects the two classes to train the SVM
              if winRep >= length(prevArr)
                  i = prevArr(winRep) ;       
                  j = prevArr(winRep-1) ;   
              else
                  i = prevArr(winRep)  ;      
                  j = prevArr(winRep+1)  ; 
              end

              % Selects the features of the 2 classes
              feature = [Wf(1:2,n*i-(n-1):n*i),Wf(1:2,n*j-(n-1):n*j)]';

              % Assigns the labels for each class
              for m1 = 1:n
                 label(m1) = 1; 
              end

              for n1 = n+1:2*n
                 label(n1) = -1; 
              end

              % The SVM is trained
              model=svmtrain(label', feature, params);
              % model = svmtrain(data(tr,:),groups(tr),'showplot',false,'kernel_function','linear');

              % The face that the user selected is classified to any of the two
              % classes
              guessLab(1) = 1;
              predLabel=svmpredict(guessLab',[Wrec(1) Wrec(2)],model);

              predLabel;

              % A winner class is selected
              if predLabel == 1 
                  winnerArr = [winnerArr i];
              elseif predLabel == -1 
                  winnerArr = [winnerArr j];
              end

              if winnerArr > 1
                  for c1 = 2:length(winnerArr)
                      if winnerArr(c1) == winnerArr(c1-1)
                         winnerArr(c1) = []; 
                      end
                  end
              end

          end

          prevArr = winnerArr;

         if length(winnerArr) < 2
             winnerArr;
              break
          end
      end

      % This is the class that was selected
      CLASS = winnerArr;

      display('The class that matches your face is:')
      display(CLASS)

      % Plot the selected face
      figure(1)
      imagesc(reshape(Tr,nRow,nCol));
      colormap gray;
      title('Face selected')

      % Plot the faces of the matching class
      for i = 1:n
          figure(i+1)
          imagesc(reshape(T(:,winnerArr*n-i+1),nRow,nCol));
          colormap gray;
          title(strcat('Matching face number: ',num2str(winnerArr*n-i+1)));
      end
  end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% SELECT FACE BASED ON THE SHORTEST EUCLIDEAN DISTANCE
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

if classType == 2

      % Calculate euclidean distance
      %testNum = 134;

      % Normalize selected image
      Tr = reshape(face(:,:,testNum),[nRow*nCol 1]);
      Ar = Tr-mTot;

      % Obtain weights of the selected face
      Wrec = Vf'*U'*Ar;

      temp = 0;

      % Obtaining an array of euclidean distances to each face
      eDist = [];
      for i = 1:M
          eDist = [eDist sqrt(( norm( Wrec - Wf(:,i)) )^2)]; 
      end

      % Find minimum distance and the corresponding index
      minDis = 999999;
      minIndex = 0;

      for i = 1:length(eDist)
         if minDis > eDist(i) && i ~= testNum
             minDis = eDist(i);
             minIndex = i;
         end
      end

      Matching_index = minIndex;

      % Matching index
      display(Matching_index);

      % Plot selected face
      figure(1)
      imagesc(reshape(Tr,nRow,nCol));
      colormap gray;
      title('Face selected')

      % Plot best match
      figure(2)
      imagesc(reshape(T(:,minIndex),nRow,nCol));
      colormap gray;
      title('Best match')
  end

This is my code, when i ran this code while seclecting svm option

it gives this error

Please select your dataset: 
(1) simple.m 
(2) pose.m
1
Enter the number of the face for which you want to find the match: 5
Choose (1) for SVM or (2) for Euclidean Distance: 1
Loading please wait...
Error using svmtrain (line 234)
Y must be a vector or a character array.

Error in svmface (line 158)
            model=svmtrain(label', feature, params);

[Sajja Tulasi Krishna]

this is database i used

[Sajja Tulasi Krishna]

data

[Steven Lord]

I have not read through the whole block of code you posted but the error message suggests to me that svmtrain is working correctly, but you're calling it with data that doesn't meet its expectations.

Set a breakpoint on line 158 of svmface and run this code. When MATLAB reaches the breakpoint, look at the size and class of the variable feature. svmtrain expects it to be a vector or a char array, and it is not. You will need to work your way backwards through the svmface code to determine why that variable is neither a vector nor a char array and how to correct that problem.

[Sajja Tulasi Krishna]

there is no y variable in this code

[Sajja Tulasi Krishna]

what is y in it i don't know please solve this error

Answer 2

[farid jafary]

Hi, does anyone have the MatLab PCA-LDA Code MATLAB code?