Multi Class Classification with "Bad" Class


I’m using multi class classification but for some instances choosing one class is a “bad” decision than other (e.g., there are weights per-instance-per class for “how bad” is this class).
I’m currently using cross entropy loss, is there a better loss function for it?


Hi Hadar!

If I understand your use case correctly, you might find some ideas about
how to address your issue in this post:


K. Frank