no code implementations • 25 Jun 2020 • Justin Lizama
A new loss function is proposed for neural networks on classification tasks which extends the hinge loss by assigning gradients to its critical points.
