"""
Created on Thu Dec 21 16:22:56 2017
@author: Jiajie
"""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import torch
import torch.nn.functional as F
from torch.autograd import Variable
def bootstrap_cross_entropy(input, target, ishard=False, beta=0.95, weight=None, size_average=True):
r"""Function that measures Cross Entropy between target and output
logits with prediction consistency(bootstrap)
Args:
input: Variable of arbitrary shape
target: Variable :math:`(N)` where each value is
`0 <= targets[i] <= C-1
ishard: Choose soft/hard bootstrap mode
beta: Weight between ``gt`` label and prediction. In paper, 0.8 for hard and 0.95 for soft
weight (Variable, optional): a manual rescaling weight
if provided it's repeated to match input tensor shape
size_average (bool, optional): By default, the losses are averaged
over observations for each minibatch. However, if the field
sizeAverage is set to False, the losses are instead summed
for each minibatch. Default: ``True``
Examples::
>>> input = autograd.Variable(torch.randn(3, 5), requires_grad=True)
>>> target = autograd.Variable(torch.LongTensor(3).random_(5))
>>> loss = bootstrap_cross_entropy(input, target)
>>> loss.backward()
"""
input_prob = F.softmax(input)
target_onehot = Variable(input.data.new(input.data.size()).zero_())
target_onehot.scatter_(1, target.view(-1,1), 1)
# print(target_onehot)
if ishard:
_,idx = input_prob.max(1)
target_onehot = target_onehot * beta + \
Variable(input.data.new(input.data.size()).zero_()).scatter_(1, idx.view(-1,1), 1) * (1-beta)
else:
target_onehot = target_onehot * beta + input_prob * (1-beta)
loss = - target_onehot * F.log_softmax(input)
#print(loss.size())
#print(weight.size())
#if weight is not None:
# loss = loss.sum(1) * weight
if size_average:
if weight is not None:
return (loss.sum(1) * weight).mean()
return loss.sum(1).mean()
else:
return loss.sum()
def BCE_bootstrap_with_logits(input, target, ishard=False, beta=0.95, weight=None, size_average=True):
r"""Function that measures Binary Cross Entropy between target and output
logits with prediction consistency(bootstrap)
Args:
input: Variable of arbitrary shape
target: Variable of the same shape as input
ishard: Choose soft/hard bootstrap mode
beta: Weight between ``gt`` label and prediction. In paper, 0.8 for hard and 0.95 for soft
weight (Variable, optional): a manual rescaling weight
if provided it's repeated to match input tensor shape
size_average (bool, optional): By default, the losses are averaged
over observations for each minibatch. However, if the field
sizeAverage is set to False, the losses are instead summed
for each minibatch. Default: ``True``
Examples::
>>> input = autograd.Variable(torch.randn(3), requires_grad=True)
>>> target = autograd.Variable(torch.FloatTensor(3).random_(2))
>>> loss = BCE_bootstrap_with_logits(input, target)
>>> loss.backward()
"""
if not (target.size() == input.size()):
raise ValueError("Target size ({}) must be the same as input size ({})".format(target.size(), input.size()))
input_prob = torch.sigmoid(input)
if ishard:
target = target * beta + (input_prob>0.5) * (1-beta)
else:
target = target * beta + input_prob * (1-beta)
print(target)
max_val = (-input).clamp(min=0)
loss = input - input * target + max_val + ((-max_val).exp() + (-input - max_val).exp()).log()
if weight is not None:
loss = loss * weight
if size_average:
return loss.mean()
else:
return loss.sum()
