tf.keras.metrics.kl_divergence

Computes Kullback-Leibler divergence loss between y_true & y_pred.

View aliases

Main aliases

tf.keras.losses.KLD, tf.keras.losses.kl_divergence, tf.keras.losses.kld, tf.keras.losses.kullback_leibler_divergence, tf.keras.metrics.KLD, tf.keras.metrics.kld, tf.keras.metrics.kullback_leibler_divergence, tf.losses.KLD, tf.losses.kl_divergence, tf.losses.kld, tf.losses.kullback_leibler_divergence, tf.metrics.KLD, tf.metrics.kl_divergence, tf.metrics.kld, tf.metrics.kullback_leibler_divergence

Compat aliases for migration

See Migration guide for more details.

`tf.compat.v1.keras.losses.KLD`, `tf.compat.v1.keras.losses.kl_divergence`, `tf.compat.v1.keras.losses.kld`, `tf.compat.v1.keras.losses.kullback_leibler_divergence`, `tf.compat.v1.keras.metrics.KLD`, `tf.compat.v1.keras.metrics.kl_divergence`, `tf.compat.v1.keras.metrics.kld`, `tf.compat.v1.keras.metrics.kullback_leibler_divergence`

tf.keras.metrics.kl_divergence(
    y_true, y_pred
)

loss = y_true * log(y_true / y_pred)

See: https://en.wikipedia.org/wiki/Kullback%E2%80%93Leibler_divergence

Standalone usage:

y_true = np.random.randint(0, 2, size=(2, 3)).astype(np.float64)
y_pred = np.random.random(size=(2, 3))
loss = tf.keras.losses.kullback_leibler_divergence(y_true, y_pred)
assert loss.shape == (2,)
y_true = tf.keras.backend.clip(y_true, 1e-7, 1)
y_pred = tf.keras.backend.clip(y_pred, 1e-7, 1)
assert np.array_equal(
    loss.numpy(), np.sum(y_true * np.log(y_true / y_pred), axis=-1))

Args
y_true	Tensor of true targets.
y_pred	Tensor of predicted targets.

Returns
A Tensor with loss.

Raises
TypeError	If y_true cannot be cast to the y_pred.dtype.