View source on GitHub |
Binary crossentropy between an output tensor and a target tensor.
tf.keras.backend.binary_crossentropy(
target, output, from_logits=False
)
target
: A tensor with the same shape as output
.output
: A tensor.from_logits
: Whether output
is expected to be a logits tensor.
By default, we consider that output
encodes a probability distribution.A tensor.