DECLoss

class torch_uncertainty.losses.DECLoss(annealing_step=None, reg_weight=None, loss_type='log', reduction='mean')

The deep evidential classification loss.

Parameters:

• annealing_step (int) – Annealing step for the weight of the

• term. (regularization) –

• reg_weight (float) – Fixed weight of the regularization term.

• loss_type (str, optional) – Specifies the loss type to apply to the

• parameters (Dirichlet) – 'mse' | 'log' | 'digamma'.

• reduction (str, optional) – Specifies the reduction to apply to the

• output – 'none' | 'mean' | 'sum'.

Reference:

Sensoy, M., Kaplan, L., & Kandemir, M. (2018). Evidential deep learning to quantify classification uncertainty. NeurIPS 2018. https://arxiv.org/abs/1806.01768.