Is it possible to use a neural network(or another approach) to classify image based on trained data and at the same time if new image classes are introduced in the test set it should classify those unseen images(open set data) to new classes(kind of telling me which new class this new unseen data belongs to?) on which training is not done.
3
-
$\begingroup$ I think that's not what open set classification means: yes it means that there is a possibility that an instance doesn't belong to any of the known classes, but such an instance is just labelled as "unknown class" or "other". As far as I know there's no way for a (supervised) classification system to identify a new class (or to even represent it). $\endgroup$– ErwanJan 23, 2021 at 23:11
-
$\begingroup$ Thanks for comment. So currently we can just get output as Unknown/unseen for open set. But can we combine zero shot learning to classify them(unknown class)? $\endgroup$– Rambo_johnJan 26, 2021 at 10:04
-
$\begingroup$ Maybe, I don't know much about zero-shot learning. As far as I understand it relies on representing classes themselves in some vector space, as opposed to regular categorical classes. $\endgroup$– ErwanJan 26, 2021 at 11:47
Add a comment
|