CODEWITHSUNDEEP
pythonmachine-learningsvmtext-classification
Manasvini Ganesh
Manasvini Ganesh

Reputation: 523

Classifying data into 3 classes with a binary classifier model

Hi I am new to Machine Learning. I have a task of classifying data into 3 classes. I have data for only 2 classes. I have used SVM Linear Classifier for the classes for which I have data (2 classes say A and B). Now there is another third class C for which I don't have any data. is there a way that I can use this binary classifier and predict if a sample doesn't belong to any of the 2 classes(A and B) then it must belong to C ?

Upvotes: 0

Views: 941

Answers (1)

radpet
radpet

Reputation: 731

You can indeed eventually use your model by observing if a specific threshold (activation) is met or not but I am not sure what will be good in practice. I doubt you can find a stable threshold values that will successfully divide the dimension so you can classify class C. I would rather try another approach. I would concat A and B and train an outlier detector such as One Class SVM. Then you would be able to classify C i.e it is neither A nor B .

Upvotes: 2

Related Questions