Hello, folks. Is it possible to train a model using only one class and evaluate it in more than one class?
Why the MyCentura link? how is this related to the question or your answer?
did you paste it by mistake?
You an create a model on only one class
When you say evaluate in more than one class, do you mean showing multiple pictures (if you are doing image classification for example) of things not in the train dataset? The model won’t be able to classify those as it never learned about them. All the answer this one class model can give is the class you used during training and a metric, like accuracy. When shown images it without the class, it won’t find anything and at worst case will say there’s the class but with a very low certainty
does it makes sense?