def testTwoClassLogLikelihoodVersusOldImplementation(self):
def alt_two_class_log_likelihood_impl(predictions, labels):
float_labels = tf.cast(labels, dtype=tf.float64)
float_predictions = tf.cast(tf.squeeze(predictions),
dtype=tf.float64)
# likelihood should be just p for class 1, and 1 - p for class 0.
# signs is 1 for class 1, and -1 for class 0
signs = 2 * float_labels - tf.ones_like(float_labels)
# constant_term is 1 for class 0, and 0 for class 1.
constant_term = tf.ones_like(float_labels) - float_labels
likelihoods = constant_term + signs * float_predictions
log_likelihoods = tf.log(likelihoods)
avg_log_likelihood = tf.reduce_mean(log_likelihoods)
return avg_log_likelihood
predictions = np.random.rand(1, 10, 1)
targets = np.random.randint(2, size=10)
with self.test_session() as session:
new_log_likelihood, _ = metrics.two_class_log_likelihood(
predictions, targets)
alt_log_likelihood = alt_two_class_log_likelihood_impl(
predictions, targets)
new_impl, alt_impl = session.run(
[new_log_likelihood, alt_log_likelihood])
self.assertAlmostEqual(new_impl, alt_impl)
def testTwoClassLogLikelihood(self):
predictions = np.array([0.0, 0.2, 0.4, 0.6, 0.8, 1.0])
targets = np.array([0, 0, 1, 0, 1, 1])
expected = (2.0 * np.log(0.8) + 2.0 * np.log(0.4)) / 6.0
with self.test_session() as session:
avg_log_likelihood, _ = metrics.two_class_log_likelihood(
predictions, targets)
actual = session.run(avg_log_likelihood)
self.assertAlmostEqual(actual, expected)