A simple squared-error reformulation for ordinal classification

Christopher Beckham, Christopher Pal

In this paper, we explore ordinal classification (in the context of deep neural networks) through a simple modification of the squared error loss which not only allows it to not on…