MATLAB: Create target for neural network

lbpneural network

Hi all,

I had extracted feature vector of an image and saved it in a excel document. feature vector is 42×42 dimension. I don't know how to create target for this input so i can train the neural network. I need to do emotion classification. 5 classes. So, how do i create target vector and train the network?

Best Answer

Hi, Remo

Feature vector that extracted from an image ideally m x 1 matrix.

You said that it's 42 x 42. Is it means 42 feature of 42 images??

If so, then you need to create 5 x 42 matrix of target.

First create matrix of zero 5 x 42.

for feature#1, this feature belong to which emotion? Let say 3rd class. Then you need to mark at first column 3rd row as 1.

t = [0 0 0 0 0 ... until 42
     0 0 0 0 0 ...
     1 0 0 0 0 ...
     0 0 0 0 0 ...
     0 0 0 0 0 ...

And so on.

I have an example below. In this example I have 5x5 feature vector. and 5x5 matrix of target.

p = [1  0  1  0  1
     0  1  1  0  1
     1  0  0  1  1
     0  1  0  1  1
     1  0  0  1  1];
t = [0 0 1 0 0
     0 0 0 0 1
     1 0 0 0 0
     0 1 0 0 0
     0 0 0 1 0];
PR = [0 1;0 1;0 1;0 1;0 1];
net = newff(PR,[5 25 25 5],{'logsig','logsig','logsig','logsig'},'traingda');
net.trainParam.epochs = 1500;
net.trainParam.goal = 0;
net = train(net,p,t);

Then try to simulate the first feature.

sim(net,p(:,1))

And the result :

It means this inputs belong to 3rd class.

I hope this will helps.

Related Solutions

MATLAB: I m working with JAFFE dataset.how to divide dataset into training and testing for neural network i m giving LBP feature of JAFFE images as a input.i have a code which divide 70% -30% ratio but how can i know that this 70% is for training,30%testin

As the above code indicates, the default is 70/15/15 . To change it, just change the percentages in the above assignment statements. HOWEVER, I do not recommend changing it !!!

If you use 70/0/30, you are flirting with the potentially disastrous possibility of overtraining an overfit net (i.e., No. of unknown weights, and thresholds, Nw, exceeds the number of training equations Ntrneq)

To mitigate this without a validation set for a regression/curve-fitting net like FITNET, use MSEREG instead of MSE and/or TRAINBR instead of TRAINLM and/or No. of training equations Ntrneq >> No. of unknown weights and biases Nw

HOWEVER, it is not clear to me what to do for a classification/pattern-recognition net like PATTERNNET that uses CROSSENTROPY instead of MSE and TRAINSCG instead of TRAINLM.

   [ I N ] = size(input)
   [ O N ] = size(target)  
 Ntrneq = Ntrn*O 
 Nw      = (I+1)*H+(H+1)*O

Obviously want

 H << (Ntrn*O -O)/ (I+O+1)

Need size(input) to continue for overfitting check

Hope this helps

Thank you for formally accepting my answer

Greg

MATLAB: Backpropogation ,Multilayer perceptron,neural network

% function TrainingNet

% load Feature.txt; %load the features

% FeatureS = Feature'; %Convert to column array

% load Outtype.txt; %load output type

% OuttypeS = Outtype';

[I N ] = size(FeatureS)

[O N ] = size(Outtypes)

minmaxF = minmax(FeatureS) % Is a matrix [I 2]

Neq = N*O % Number of training equations

% I-H-O node topology

% Nw = (I+1)*H+(H+1)*O % Number of unknown weights

% Want Neq >> Nw or % H << Hub

Hub = (Neq-O)/(I+O+1) % Neq = Nw

r = 10 % Neq > r*Nw, ~2 < r < ~30

H = floor((Neq/r-O)/(I+O+1))

How did you get H = 2000 ???

% %initialize parameters for creating the MLP.

% fcnCELL = {'logsig' 'logsig'};

% initflag = [0 1];

What does initflag do?

% trainalgo = 'gdm';

% paramatrix = [10000 50 0.9 0.6]; % epochs = 100, show = 50,

100 or 10,000?

                                   % learning rate = 0.9, momentum term = 0.6

% sameWEIGHT = [];

I suggest first using the defaults in NEWFF

% net_FFBP = creteNet(inputsize, mimax, hneurons, fcnCELL, initflag, trainalgo, paramatrix, sameWEIGHT);

Is this supposed to be a replacement for NEWFF and net.Param.* ??

% net_FFBP = newff(FeatureS, OuttypeS, 39);

Now H = 39 ??

% [net_FFBP] = train(net_FFBP, FeatureS, OuttypeS);

% save net_FFBP net_FFBP;

% disp('Done: Training Network');

What is your question ??

Greg

Best Answer

Related Solutions

MATLAB: I m working with JAFFE dataset.how to divide dataset into training and testing for neural network i m giving LBP feature of JAFFE images as a input.i have a code which divide 70% -30% ratio but how can i know that this 70% is for training,30%testin

MATLAB: Backpropogation ,Multilayer perceptron,neural network

Related Question