How to train SVM

Hi, Is there anyone who will help me in SVM for classification in Matlab code. i have completed my work until feature extraction and after feature extraction, I have created mxn size of the matrix where n is the number of samples and mx1 is the array of each image/character.

Please guide that, how I have to arrange the training dataset that I can train SVM.

Answer 1

[Walter Roberson]

https://www.mathworks.com/help/stats/svmtrain.html

"Input Arguments:

Train: Matrix of training data, where each row corresponds to an observation or replicate, and each column corresponds to a feature or variable."

That is already the form that your indicate your data is in, so you do not need to do anything further to prepare it. Just call

svmtrain(YourMatrix, Vector_of_target_information)

[Ameer Hamza]

svmtrain() has been removed from R2018a. Now alternative is fitcsvm().

[neha gautam]

I tried to use your code. but i got this error

[neha gautam]

Sorry to say, I have 2017b and after that, I am getting other errors. Please guide that how can I solve it.

[Walter Roberson]

What is class(tar) and size(tar) ?

[neha gautam]

My training dataset is 26x2600 and target is 1x2600, I have solved the previous error. now I get a new one.

Answer 2

[Ameer Hamza]

The easiest way to get started and visualize is to use Classification Learner App. Start it using this command

classificationLearner

Then you can start a new session. Import your data. In the model type choose SVM. Several SVM models are available. Choose advanced and choose advanced training options. Then press train to start training. When Training is complete, it will visualize result using several graphs. You can also export the model to the workspace for further processing.

Answer 3

[Sindhu]

How to train SVM in satellite images?