Can Borderline SMOTE be used for multiple categories?

The terms "positive data points" and "negative data points" lead me to believe that the example in this video is referring to binary classification. Can Borderline SMOTE still be used when there are multiple categories in a feature?  Thank you.