def test_fc_with_dropout(self):
with tf.Graph().as_default(), tf.Session() as sess:
tf.set_random_seed(10)
x_ = tf.ones(shape=[3, 24], dtype=tf.float32)
with tf.variable_scope("Hidden"):
h_ = models.fully_connected_layers(x_, [100],
dropout_rate=0.5,
is_training=False)
h_nz_ = tf.reduce_sum(tf.cast(tf.equal(h_, 0), dtype=tf.int32))
with tf.variable_scope("Hidden", reuse=True):
h_train_ = models.fully_connected_layers(x_, [100],
dropout_rate=0.5,
is_training=True)
h_train_nz_ = tf.reduce_sum(
tf.cast(tf.equal(h_train_, 0), dtype=tf.int32))
self.assertEqual(h_.get_shape().as_list(), [3, 100])
self.assertEqual(h_train_.get_shape().as_list(), [3, 100])
sess.run(tf.global_variables_initializer())
h_nz, h_train_nz = sess.run([h_nz_, h_train_nz_])
self.assertEqual(h_nz, 0)
# Check that with dropout, number of zeros is close to expected
expected_num_zero = 0.5 * np.prod(h_.get_shape().as_list())
self.assertLessEqual(h_train_nz, expected_num_zero * 1.1)
self.assertGreaterEqual(h_train_nz, expected_num_zero * 0.9)
def _build_network(self, dim_a, params):
with tf.variable_scope('base'):
# Initialize graph
if params['train_ae']:
ae_trainer = TrainAE()
ae_trainer.run(train=True, show_performance=True)
self.AE = AE(ae_loc=params['ae_loc'])
ae_encoder = self.AE.latent_space
self.low_dim_input_shape = ae_encoder.get_shape()[1:]
self.low_dim_input = tf.placeholder(tf.float32, [
None, self.low_dim_input_shape[0], self.low_dim_input_shape[1],
self.low_dim_input_shape[2]
],
name='input')
self.low_dim_input = tf.identity(self.low_dim_input,
name='low_dim_input')
# Build fully connected layers
self.y, loss = fully_connected_layers(
tf.contrib.layers.flatten(self.low_dim_input), dim_a,
params['fc_layers_neurons'], params['loss_function_type'])
variables = tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES, 'base')
self.train_policy = tf.train.GradientDescentOptimizer(
learning_rate=params['learning_rate']).minimize(loss,
var_list=variables)
# Initialize tensorflow
init = tf.global_variables_initializer()
self.sess = tf.Session()
self.sess.run(init)
self.saver = tf.train.Saver()
def _build_network(self, dim_a, params):
# Initialize graph
with tf.variable_scope('base'):
# Build autoencoder
ae_inputs = tf.placeholder(
tf.float32, (None, self.image_size, self.image_size, 1),
name='input')
self.loss_ae, latent_space, self.ae_output = autoencoder(ae_inputs)
# Build fully connected layers
self.y, loss_policy = fully_connected_layers(
tf.contrib.layers.flatten(latent_space), dim_a,
params['fc_layers_neurons'], params['loss_function_type'])
variables = tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES, 'base')
self.train_policy = tf.train.GradientDescentOptimizer(
learning_rate=params['learning_rate']).minimize(loss_policy,
var_list=variables)
self.train_ae = tf.train.AdamOptimizer(
learning_rate=params['learning_rate']).minimize(self.loss_ae)
# Initialize tensorflow
init = tf.global_variables_initializer()
self.sess = tf.Session()
self.sess.run(init)
self.saver = tf.train.Saver()
def test_fc_with_dropout(self):
with tf.Graph().as_default(), tf.Session() as sess:
tf.set_random_seed(10)
x_ = tf.ones(shape=[3, 24], dtype=tf.float32)
with tf.variable_scope("Hidden"):
h_ = models.fully_connected_layers(x_, [100], dropout_rate=0.5,
is_training=False)
h_nz_ = tf.reduce_sum(tf.cast(tf.equal(h_, 0),
dtype=tf.int32))
with tf.variable_scope("Hidden", reuse=True):
h_train_ = models.fully_connected_layers(x_, [100], dropout_rate=0.5,
is_training=True)
h_train_nz_ = tf.reduce_sum(tf.cast(tf.equal(h_train_, 0),
dtype=tf.int32))
self.assertEqual(h_.get_shape().as_list(), [3, 100])
self.assertEqual(h_train_.get_shape().as_list(), [3, 100])
sess.run(tf.global_variables_initializer())
h_nz, h_train_nz = sess.run([h_nz_, h_train_nz_])
self.assertEqual(h_nz, 0)
# Check that with dropout, number of zeros is close to expected
expected_num_zero = 0.5 * np.prod(h_.get_shape().as_list())
self.assertLessEqual(h_train_nz, expected_num_zero * 1.1)
self.assertGreaterEqual(h_train_nz, expected_num_zero * 0.9)
def _build_network(self, dim_a, params):
with tf.variable_scope('base'):
# Input data
x = tf.placeholder(tf.float32,
[None, params['low_dim_input_shape']],
name='input')
# Build fully connected layers
self.y, loss = fully_connected_layers(x, dim_a,
params['fc_layers_neurons'],
params['loss_function_type'])
variables = tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES, 'base')
self.train_step = tf.train.GradientDescentOptimizer(
learning_rate=params['learning_rate']).minimize(loss,
var_list=variables)
# Initialize tensorflow
init = tf.global_variables_initializer()
self.sess = tf.Session()
self.sess.run(init)
self.saver = tf.train.Saver()