def test_missing_blanket(self):
N = 10
z = rvs.Bernoulli(p=0.75, sample_shape=N)
z_cond = ed.complete_conditional(z)
self.assertIsInstance(z_cond, rvs.Bernoulli)
with self.test_session() as sess:
p_val = sess.run(z_cond.p)
self.assertAllClose(p_val, 0.75 + np.zeros(N, np.float32))
def test_basic_bernoulli(self):
N = 10
z = rvs.Bernoulli(probs=0.75, sample_shape=N)
z_cond = ed.complete_conditional(z, [z])
self.assertIsInstance(z_cond, rvs.Bernoulli)
with self.test_session() as sess:
p_val = sess.run(z_cond.probs)
self.assertAllClose(p_val, 0.75 + np.zeros(N, np.float32))
def test_beta_bernoulli(self):
x_data = np.array([0, 1, 0, 0, 0, 0, 0, 0, 0, 1])
a0 = 0.5
b0 = 1.5
pi = rvs.Beta(a=a0, b=b0)
x = rvs.Bernoulli(p=pi, sample_shape=10)
pi_cond = ed.complete_conditional(pi, [pi, x])
self.assertIsInstance(pi_cond, rvs.Beta)
with self.test_session() as sess:
a_val, b_val = sess.run([pi_cond.a, pi_cond.b], {x: x_data})
self.assertAllClose(a_val, a0 + x_data.sum())
self.assertAllClose(b_val, b0 + (1 - x_data).sum())
For Part II, we would like you to hand in your code, and the runs that produce the two answers"""
import tensorflow as tf
print("tensorflow version: %s" % tf.__version__)
import edward as ed
print("edward version: %s" % ed.__version__)
import edward.models as edm
import edward.inferences as edi
matplotlib inline
config InlineBackend.figure_format = 'retina'
import matplotlib.pyplot as plt
rain = edm.Bernoulli(probs=0.2)
p_sprinkler = tf.where(tf.cast(rain, tf.bool), 0.01, 0.4)
sprinkler = edm.Bernoulli(probs=p_sprinkler)
p_grass_wet = tf.where(tf.cast(rain, tf.bool),
tf.where(tf.cast(sprinkler, tf.bool), 0.99, 0.8),
tf.where(tf.cast(sprinkler, tf.bool), 0.9, 0.00000001))
grass_wet = edm.Bernoulli(probs=p_grass_wet)
with tf.Session():
plt.hist([grass_wet.eval() for _ in range(1000)]);
q_rain = edm.Bernoulli(probs=tf.nn.sigmoid(tf.Variable(tf.random_normal([]))))
ed.get_session()
inf = edi.KLpq({rain: q_rain}, data={grass_wet: tf.constant(1, dtype=tf.int32)})
inf.run(n_samples=50)
print(q_rain.probs.eval())