def test_cce_one_hot():
y_a = K.variable(np.random.randint(0, 7, (5, 6)))
y_b = K.variable(np.random.random((5, 6, 7)))
objective_output = objectives.sparse_categorical_crossentropy(y_a, y_b)
assert K.eval(objective_output).shape == (5, 6)
y_a = K.variable(np.random.randint(0, 7, (6,)))
y_b = K.variable(np.random.random((6, 7)))
assert K.eval(objectives.sparse_categorical_crossentropy(y_a, y_b)).shape == (6,)
def test_cce_one_hot():
y_a = K.variable(np.random.randint(0, 7, (5, 6)))
y_b = K.variable(np.random.random((5, 6, 7)))
objective_output = objectives.sparse_categorical_crossentropy(y_a, y_b)
assert K.eval(objective_output).shape == (5, 6)
y_a = K.variable(np.random.randint(0, 7, (6, )))
y_b = K.variable(np.random.random((6, 7)))
assert K.eval(objectives.sparse_categorical_crossentropy(
y_a, y_b)).shape == (6, )
def multi_sparse_graph_loss(y_true, y_pred):
ids = tensor.nonzero(y_true[:, 0] + 1)[0]
y_true = y_true[ids]
y_pred = y_pred[ids]
return tensor.mean(
objectives.sparse_categorical_crossentropy(y_true, y_pred))
def vae_loss(self, x, x_decoded_onehot):
xent_loss = objectives.sparse_categorical_crossentropy(
x, x_decoded_onehot)
kl_loss = -0.5 * K.mean(1 + self.z_log_sigma - K.square(self.z_mean) -
K.exp(self.z_log_sigma))
loss = xent_loss + self.kl_w * kl_loss
return loss
def logp(self, a):
logps = []
for i, (low, high) in enumerate(
zip(self.action_space.low, self.action_space.high)):
if low + 1 == high:
logp = (a[:, :, :1] * K.log(self.ps[i]) +
(1 - a[:, :, :1]) * K.log(1 - self.ps[i]))[:, :, 0]
a = a[:, :, 1:]
else:
logp = -sparse_categorical_crossentropy(
a[:, :, :1], self.ps[i])
a = a[:, :, 1:]
logps.append(logp)
return maybe_merge(logps, mode="sum"), a
def crf_loss(y_true, y_pred):
"""General CRF loss function depending on the learning mode.
# Arguments
y_true: tensor with true targets.
y_pred: tensor with predicted targets.
# Returns
If the CRF layer is being trained in the join mode, returns the negative
log-likelihood. Otherwise returns the categorical crossentropy implemented
by the underlying Keras backend.
"""
crf, idx = y_pred._keras_history[:2]
if crf.learn_mode == 'join':
return crf_nll(y_true, y_pred)
else:
if crf.sparse_target:
return sparse_categorical_crossentropy(y_true, y_pred)
else:
return categorical_crossentropy(y_true, y_pred)
def logp(self, a):
return -sparse_categorical_crossentropy(a[:, :, :1], self.p), a[:, :,
1:]