loss.py

import numpy as np
import torch
import torch.nn as nn
from torch.autograd import Variable
import math
import torch.nn.functional as F
import pdb


class CrossEntropyLabelSmooth(nn.Module):
    """Cross entropy loss with label smoothing regularizer.
    Reference:
    Szegedy et al. Rethinking the Inception Architecture for Computer Vision. CVPR 2016.
    Equation: y = (1 - epsilon) * y + epsilon / K.
    Args:
        num_classes (int): number of classes.
        epsilon (float): weight.
    """

    def __init__(self, num_classes, epsilon=0.1, use_gpu=True, size_average=True):
        super(CrossEntropyLabelSmooth, self).__init__()
        self.num_classes = num_classes
        self.epsilon = epsilon
        self.use_gpu = use_gpu
        self.size_average = size_average
        self.logsoftmax = nn.LogSoftmax(dim=1)

    def forward(self, inputs, targets):
        """
        Args:
            inputs: prediction matrix (before softmax) with shape (batch_size, num_classes)
            targets: ground truth labels with shape (num_classes)
        """
        log_probs = self.logsoftmax(inputs)
        targets = torch.zeros(log_probs.size()).scatter_(1, targets.unsqueeze(1).cpu(), 1)
        if self.use_gpu: targets = targets.cuda()
        targets = (1 - self.epsilon) * targets + self.epsilon / self.num_classes
        if self.size_average:
            loss = (- targets * log_probs).mean(0).sum()
        else:
            loss = (- targets * log_probs).sum(1)
        return loss

    

def SKL(out1, out2):
    out2_t = out2.clone()
    out2_t = out2_t.detach()
    out1_t = out1.clone()
    out1_t = out1_t.detach()
    return (F.kl_div(F.log_softmax(out1), out2_t, reduction='none') +
            F.kl_div(F.log_softmax(out2), out1_t, reduction='none')) / 2