Large Margin Deep Networks For Classification NeurIPS 2018

https://arxiv.org/abs/1803.05598

Work in progress

This is the directory for the large margin deep network example experiments. The directory includes the large margin loss function and training/evaluation scripts using the large margin loss function. Benefits of adding large margin loss function to training includes generalization from small training sets, corrupted labels, and robustness against adversarial perturbations. See paper for more details.