def my_loss(y_true, y_pred):
entropy = weighted_cross_entropy_with_logits(
y_true, y_pred, K.constant(loss_weights, dtype='float32'))
mask, count_unmasked = mask_from(y_true, mask_val)
masked_entropy = entropy * mask
loss = K.sum(masked_entropy) / count_unmasked
return loss
def testConstructionNamed(self):
with self.test_session():
logits, targets, pos_weight, _ = self._Inputs()
loss = nn_impl.weighted_cross_entropy_with_logits(targets,
logits,
pos_weight,
name="mybce")
self.assertEqual("mybce", loss.op.name)
def testGradient(self):
sizes = [4, 2]
with self.cached_session():
logits, targets, pos_weight, _ = self._Inputs(sizes=sizes)
loss = nn_impl.weighted_cross_entropy_with_logits(
targets=targets, logits=logits, pos_weight=pos_weight)
err = gradient_checker.compute_gradient_error(logits, sizes, loss, sizes)
print("logistic loss gradient err = ", err)
self.assertLess(err, 1e-7)
def testOutput(self):
for use_gpu in [True, False]:
with self.test_session(use_gpu=use_gpu):
logits, targets, pos_weight, losses = self._Inputs(dtype=dtypes.float32)
loss = nn_impl.weighted_cross_entropy_with_logits(
targets=targets, logits=logits, pos_weight=pos_weight)
np_loss = np.array(losses).astype(np.float32)
tf_loss = loss.eval()
self.assertAllClose(np_loss, tf_loss, atol=0.001)
def testShapeError(self):
with self.assertRaisesRegexp(ValueError, "must have the same shape"):
nn_impl.weighted_cross_entropy_with_logits(
targets=[1, 2, 3], logits=[[2, 1]], pos_weight=2.0)
def testConstructionNamed(self):
with self.cached_session():
logits, targets, pos_weight, _ = self._Inputs()
loss = nn_impl.weighted_cross_entropy_with_logits(
targets=targets, logits=logits, pos_weight=pos_weight, name="mybce")
self.assertEqual("mybce", loss.op.name)
