Multiclass classification with junk class

Question

I would like to do multiclass classification but I'm in a weird spot, I'm basically trying to classify code as good or bad where there are multiple classes for bad code but there will only be one good class. It's almost like binary classification where its either good or bad but there are multiple cases for bad code. I'm wondering if making the good class the same size as all other bad classes combined will be feasible or if there is another method to go about this.

Not a *programming* question, hence off-topic here; please see the intro and NOTE in https://stackoverflow.com/tags/machine-learning/info — desertnaut, Sep 06 '22 at 14:37

score 0 · Answer 1 · answered Sep 04 '22 at 08:56

If you only care if a given code instance is good or bad, you might want to check out semi-supervised anomaly detection, where the good code is considered anomalous. It might be referred to as "Other-vs-One", which is not as common as "One-vs-Other", but regardless it is worth a shot.

Multiclass classification with junk class

1 Answers1