Doubt in labelling data for machine learning in MATLAB.

Hello all, I am trying to do classification task of Mrandom -1 and 1 symbols using SVM in MATLAB. The input feature vector is based on energy levels and is shown as and and the correspnding training labels are given as .

I am able to compute the input feature vectors but not getting how to construct the training labels.

Any help in this regard will be highly appreciated.

Answer 1

[Divyam]

Hi @charu shree,

Assuming that the SVM is used here to map the feature vectors and with their corresponding class labels and 1, the training label vector y can be created as follows:

M_minus = size(E_minus, 1);
M_plus = size(E_plus, 1);

% Create the label vector
labels_minus = -1 * ones(M_minus, 1);
labels_plus = 1 * ones(M_plus, 1);

% Combine the labels
y = [labels_minus; labels_plus];

% Combine the feature vectors
features = [E_minus; E_plus];

% Train SVM model
SVMModel = fitcsvm(features, y);

Using the attached script "SVMTest.m" you can test the code and check out the accuracy and predictions of the SVM.