def __call__(self, input):
with tf.varible_scope("decoder"):
h1 = tf.contrib.layers.conv2d_transpose(input,
num_outputs=64,
kernel_size=4,
stride=2)
h2 = tf.contrib.layers.conv2d_transpose(h1,
num_outputs=64,
kernel_size=4,
stride=2)
h3 = tf.contrib.layers.conv2d_transpose(h2,
num_outputs=32,
kernel_size=4,
stride=2)
h4 = tf.contrib.layers.conv2d_transpose(h3,
num_outputs=32,
kernel_size=4,
stride=2)
return (h4)
def __init__(self, learning_rate=0.01, scope="policy_estimator"):
with tf.varible_scope():
self.state = tf.placeholder(tf.int32, [], "state")
self.action = tf.placeholder(tf.int32, [], "action")
self.target = tf.placeholder(tf.int32, [], "target")
state_one_hot = tf.one_hot(self.state, int(env.observation_space.n))
self.output_layer = tf.contrib.layers.fullt_connected(
inputs=tf.expand_dims(state_one_hot, 0),
num_outputs=env.action_space.n,
activation_fn=None,
weights_initializer=tf.zeros.initializer)
self.action_probs = tf.squeeze(tf.nn.softmax(self.output_layer))
self.picked_Action_prob = tf.gather(self.action_probs, self.action)
self.loss = -tf.log(self.picked_Action_prob) * self.target
self.optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate)
self.train_op = self.optimizer.minimize(
self.loss, global_step=tf.contrib.framework.get_global_step())
def inception_v3_base(inputs, scope=None): # input 299*299*3
end_points = {} # 用来保存某些关键点供之后使用
with tf.variable_scope(scope, 'InceptionV3', [inputs]):
with slim.arg_scope([slim.conv2d, slim.max_pool2d, slim.avg_pool2d],
stride=1,
padding='VALID'):
net = slim.conv2d(
input, 32, [3, 3], stride=2,
scope='Conv2d_1a_3*3') # (299-3)/2+1=149 149*194*32
net = slim.conv2d(
net, 32, [3, 3],
scope='Conv2d_2a_3*3') # (149-3)/1+1=147 147*147*32
net = slim.conv2d(net,
64, [3, 3],
padding='SAME',
scope='Conv2d_3a_3*3') # 147/1 147*147*64
net = slim.max_pool2d(
net, [3, 3], stride=2,
scope='MaxPool_3a_3*3') # (147-3)/2+1=73 73*73*64
net = slim.conv2d(net, 80, [3, 3],
scope='Conv2d_3b_3*3') # (73-3)/1+1=71 71*71*80
net = slim.conv2d(net,
192, [3, 3],
stride=2,
scope='Conv2d_4a_3*3') # (71-3)/2+1=35 35*35*192
net = slim.max_pool2d(net, [3, 3],
padding='SAVE',
scope='MaxPool_5a_3*3') # 35/1 35*35*192
with slim.arg_scope([slim.conv2d, slim.max_pool2d, slim.avg_pool2d],
stride=1,
padding='SAVE'):
# 第一个模块组
with tf.variable_scope('Mixed_5b'):
with tf.variable_scope('Branch_0'):
branch_0 = slim.conv2d(net, 64, [1, 1], scope='Conv2d_0a_1*1')
with tf.variable_scope('Branch_1'):
branch_1 = slim.conv2d(net, 48, [1, 1], scope='Conv2d_0a_1*1')
branch_1 = slim.conv2d(branch_1,
64, [5, 5],
scope='Conv2d_0b_5*5')
with tf.variable_scope('Branch_3'):
branch_2 = slim.conv2d(net, 64, [1, 1], scope='Conv2d_0a_1*1')
branch_2 = slim.conv2d(branch_2,
96, [3, 3],
scope='Conv2d_0b_3*3')
branch_2 = slim.conv2d(branch_2,
96, [3, 3],
scope='Conv2d_0c_3*3')
with tf.variable_scope('Branch_3'):
branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_0a_3*3')
branch_3 = slim.conv2d(branch_3,
32, [1, 1],
scope='Conv2d_0b_1*1')
net = tf.concat([branch_0, branch_1, branch_2, branch_3], 3)
with tf.variable_scope('Mixed_5c'):
with tf.variable_scope('Branch_0'):
branch_0 = slim.conv2d(net, 64, [1, 1], scope='Conv2d_0a_1*1')
with tf.variable_scope('Branch_1'):
branch_1 = slim.conv2d(net, 48, [1, 1], scope='Conv2d_0a_1*1')
branch_1 = slim.conv2d(branch_1,
64, [5, 5],
scope='Conv2d_0b_5*5')
with tf.variable_scope('Branch_2'):
branch_2 = slim.conv2d(net, 64, [1, 1], scope='Conv2d_0a_3*3')
branch_2 = slim.conv2d(branch_2,
96, [3, 3],
scope='Conv2d_0b_3*3')
branch_2 = slim.conv2d(branch_2,
96, [3, 3],
scope='Conv2d_0c_3*3')
with tf.variable_scope('Branch_3'):
branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_0a_3*3')
branch_3 = slim.conv2d(branch_3,
64, [1, 1],
scope='Conv2d_0b_1*1')
net = tf.concat([branch_0, branch_1, branch_2, branch_3], 3)
with tf.variable_scope('Mixed_5d'):
with tf.variable_scope('Branch_0'):
branch_0 = slim.conv2d(net, 64, [1, 1], scope='Conv2d_0a_1*1')
with tf.variable_scope('Branch_1'):
branch_1 = slim.conv2d(net, 48, [1, 1], scope='Conv2d_0a_1*1')
branch_1 = slim.conv2d(branch_1,
64, [5, 5],
scope='Conv2d_0b_5*5')
with tf.variable_scope('Branch_2'):
branch_2 = slim.conv2d(net, 64, [1, 1], scope='Conv2d_0a_1*1')
branch_2 = slim.conv2d(branch_2,
96, [3, 3],
scope='Conv2d_0b_3*3')
branch_2 = slim.conv2d(branch_2,
96, [3, 3],
scope='COnv2d_0c_3*3')
with tf.varible_scope("Branch_3"):
branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_0a_3*3')
branch_3 = slim.conv2d(branch_3,
64, [1, 1],
scope="Conv2d_0b_1*1")
net = tf.concat([branch_0, branch_1, branch_2, branch_3], 3)
# 第二个模块组
with tf.variable_scope('Mixed_6a'):
with tf.variable_scope('Branch_0'):
branch_0 = slim.conv2d(net,
384, [3, 3],
stride=2,
padding='VALID',
scope='Conv2d_1a_3*3')
with tf.variable_scope('Branch_1'):
branch_1 = slim.conv2d(net, 64, [1, 1], scope='Conv2d_1a_1*1')
branch_1 = slim.conv2d(branch_1,
96, [3, 3],
scope='Conv2d_1b_3*3')
branch_1 = slim.conv2d(branch_1,
96, [3, 3],
stride=2,
padding='VALID',
scope='Conv2d_1c_3*3')
with tf.varible_scope("Branch_3"):
branch_2 = slim.max_pool2d(net, [3, 3],
stride=2,
padding='VALID',
scope='maxPool_1a_3*3')
net = tf.concat([branch_0, branch_1, branch_2], 3)
with tf.variable_scope('Mixed_6b'):
with tf.variable_scope('Branch_0'):
branch_0 = slim.conv2d(net, 192, [1, 1], scope='Conv2d_1a_1*1')
with tf.variable_scope('Branch_1'):
branch_1 = slim.conv2d(net, 128, [1, 1], scope='Conv2d_1a_1*1')
branch_1 = slim.conv2d(branch_1,
128, [1, 7],
scope='Conv2d_1b_1*7')
branch_1 = slim.conv2d(branch_1,
192, [7, 1],
scope='Conv2d_1b_7*1')
with tf.variable_scope('Branch_2'):
branch_2 = slim.conv2d(net, 128, [1, 1], scope='Conv2d_1a_1*1')
branch_2 = slim.conv2d(branch_2,
128, [7, 1],
scope='Conv2d_1b_7*1')
branch_2 = slim.conv2d(branch_2,
128, [1, 7],
scope='COnv2d_1c_1*7')
branch_2 = slim.conv2d(branch_2,
128, [7, 1],
scope='COnv2d_1c_7*1')
branch_2 = slim.conv2d(branch_2,
192, [1, 7],
scope='COnv2d_1c_1*7')
with tf.varible_scope("Branch_3"):
branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_1a_3*3')
branch_3 = slim.conv2d(branch_3,
192, [1, 1],
scope="Conv2d_1b_1*1")
net = tf.concat([branch_0, branch_1, branch_2, branch_3], 3)
with tf.variable_scope('Mixed_6c'):
with tf.variable_scope('Branch_0'):
branch_0 = slim.conv2d(net, 192, [1, 1], scope='Conv2d_1a_1*1')
with tf.variable_scope('Branch_1'):
branch_1 = slim.conv2d(net, 160, [1, 1], scope='Conv2d_1a_1*1')
branch_1 = slim.conv2d(branch_1,
160, [1, 7],
scope='Conv2d_1b_1*7')
branch_1 = slim.conv2d(branch_1,
192, [7, 1],
scope='Conv2d_1b_7*1')
with tf.variable_scope('Branch_2'):
branch_2 = slim.conv2d(net, 160, [1, 1], scope='Conv2d_1a_1*1')
branch_2 = slim.conv2d(branch_2,
160, [7, 1],
scope='Conv2d_1b_7*1')
branch_2 = slim.conv2d(branch_2,
160, [1, 7],
scope='COnv2d_1c_1*7')
branch_2 = slim.conv2d(branch_2,
160, [7, 1],
scope='COnv2d_1c_7*1')
branch_2 = slim.conv2d(branch_2,
192, [1, 7],
scope='COnv2d_1c_1*7')
with tf.varible_scope("Branch_3"):
branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_1a_3*3')
branch_3 = slim.conv2d(branch_3,
192, [1, 1],
scope="Conv2d_1b_1*1")
net = tf.concat([branch_0, branch_1, branch_2, branch_3], 3)
with tf.variable_scope('Mixed_6d'):
with tf.variable_scope('Branch_0'):
branch_0 = slim.conv2d(net, 192, [1, 1], scope='Conv2d_1a_1*1')
with tf.variable_scope('Branch_1'):
branch_1 = slim.conv2d(net, 160, [1, 1], scope='Conv2d_1a_1*1')
branch_1 = slim.conv2d(branch_1,
160, [1, 7],
scope='Conv2d_1b_1*7')
branch_1 = slim.conv2d(branch_1,
192, [7, 1],
scope='Conv2d_1b_7*1')
with tf.variable_scope('Branch_2'):
branch_2 = slim.conv2d(net, 160, [1, 1], scope='Conv2d_1a_1*1')
branch_2 = slim.conv2d(branch_2,
160, [7, 1],
scope='Conv2d_1b_7*1')
branch_2 = slim.conv2d(branch_2,
160, [1, 7],
scope='COnv2d_1c_1*7')
branch_2 = slim.conv2d(branch_2,
160, [7, 1],
scope='COnv2d_1c_7*1')
branch_2 = slim.conv2d(branch_2,
192, [1, 7],
scope='COnv2d_1c_1*7')
with tf.varible_scope("Branch_3"):
branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_1a_3*3')
branch_3 = slim.conv2d(branch_3,
192, [1, 1],
scope="Conv2d_1b_1*1")
net = tf.concat([branch_0, branch_1, branch_2, branch_3], 3)
with tf.variable_scope('Mixed_6e'):
with tf.variable_scope('Branch_0'):
branch_0 = slim.conv2d(net, 192, [1, 1], scope='Conv2d_1a_1*1')
with tf.variable_scope('Branch_1'):
branch_1 = slim.conv2d(net, 160, [1, 1], scope='Conv2d_1a_1*1')
branch_1 = slim.conv2d(branch_1,
160, [1, 7],
scope='Conv2d_1b_1*7')
branch_1 = slim.conv2d(branch_1,
192, [7, 1],
scope='Conv2d_1b_7*1')
with tf.variable_scope('Branch_2'):
branch_2 = slim.conv2d(net, 160, [1, 1], scope='Conv2d_1a_1*1')
branch_2 = slim.conv2d(branch_2,
160, [7, 1],
scope='Conv2d_1b_7*1')
branch_2 = slim.conv2d(branch_2,
160, [1, 7],
scope='COnv2d_1c_1*7')
branch_2 = slim.conv2d(branch_2,
160, [7, 1],
scope='COnv2d_1c_7*1')
branch_2 = slim.conv2d(branch_2,
192, [1, 7],
scope='COnv2d_1c_1*7')
with tf.varible_scope("Branch_3"):
branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_1a_3*3')
branch_3 = slim.conv2d(branch_3,
192, [1, 1],
scope="Conv2d_1b_1*1")
net = tf.concat([branch_0, branch_1, branch_2, branch_3], 3)
# 保留
end_points['Mixed_6e'] = net
# 第三个模块
with tf.variable_scope('Mixed_7a'):
with tf.variable_scope('Branch_0'):
branch_0 = slim.conv2d(net, 192, [1, 1], scope='Conv2d_1a_1*1')
branch_0 = slim.conv2d(branch_0,
320, [3, 3],
stride=2,
padding='VALID',
scope='Conv2d_1a_1*1')
with tf.variable_scope('Branch_1'):
branch_1 = slim.conv2d(net, 192, [1, 1], scope='Conv2d_1a_1*1')
branch_1 = slim.conv2d(branch_1,
192, [1, 7],
scope='Conv2d_1b_1*7')
branch_1 = slim.conv2d(branch_1,
192, [7, 1],
scope='Conv2d_1b_7*1')
branch_1 = slim.conv2d(branch_1,
192, [3, 3],
stride=2,
padding='VALID',
scope='Conv2d_1a_1*1')
with tf.varible_scope("Branch_2"):
branch_2 = slim.max_pool2d(net, [3, 3],
stride=2,
padding='VALID',
scope='maxPool_1a_3*3')
net = tf.concat([branch_0, branch_1, branch_2], 3)
with tf.variable_scope('Mixed_7b'):
with tf.variable_scope('Branch_0'):
branch_0 = slim.conv2d(net, 320, [1, 1], scope='Conv2d_1a_1*1')
with tf.variable_scope('Branch_1'):
branch_1 = slim.conv2d(net, 384, [1, 1], scope='Conv2d_1a_1*1')
branch_1 = tf.concat([
slim.conv2d(branch_1, 384, [1, 3], scope='Conv2d_1b_1*3'),
slim.conv2d(branch_1, 384, [3, 1], scope='Conv2d_1b_3*1')
], 3)
with tf.variable_scope('Branch_2'):
branch_2 = slim.conv2d(net, 448, [1, 1], scope='Conv2d_1a_1*1')
branch_2 = slim.conv2d(branch_2,
384, [3, 3],
scope='Conv2d_1a_1*1')
branch_2 = tf.concat([
slim.conv2d(branch_2, 384, [1, 3], scope='Conv2d_1b_1*3'),
slim.conv2d(branch_2, 384, [3, 1], scope='Conv2d_1b_3*1')
], 3)
with tf.varible_scope("Branch_3"):
branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_1a_3*3')
branch_3 = slim.conv2d(branch_3,
192, [1, 1],
scope="Conv2d_1b_1*1")
net = tf.concat([branch_0, branch_1, branch_2, branch_3], 3)
with tf.variable_scope('Mixed_7c'):
with tf.variable_scope('Branch_0'):
branch_0 = slim.conv2d(net, 320, [1, 1], scope='Conv2d_1a_1*1')
with tf.variable_scope('Branch_1'):
branch_1 = slim.conv2d(net, 384, [1, 1], scope='Conv2d_1a_1*1')
branch_1 = tf.concat([
slim.conv2d(branch_1, 384, [1, 3], scope='Conv2d_1b_1*3'),
slim.conv2d(branch_1, 384, [3, 1], scope='Conv2d_1b_3*1')
], 3)
with tf.variable_scope('Branch_2'):
branch_2 = slim.conv2d(net, 448, [1, 1], scope='Conv2d_1a_1*1')
branch_2 = slim.conv2d(branch_2,
384, [3, 3],
scope='Conv2d_1a_1*1')
branch_2 = tf.concat([
slim.conv2d(branch_2, 384, [1, 3], scope='Conv2d_1b_1*3'),
slim.conv2d(branch_2, 384, [3, 1], scope='Conv2d_1b_3*1')
], 3)
with tf.varible_scope("Branch_3"):
branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_1a_3*3')
branch_3 = slim.conv2d(branch_3,
192, [1, 1],
scope="Conv2d_1b_1*1")
net = tf.concat([branch_0, branch_1, branch_2, branch_3], 3)
return net, end_points
def generator(noise, labels):
with tf.varible_scope('generator'):
inputs = tf.concat([noise, lables], 1)
hidden = tf.layers.dense(inputs, n_hidden, activation = tf.nn.relu)
output = tf.layers.dense(hidden, n_input, activation = tf.nn.sigmoid)
return output
def __call__(self, input):
with tf.varible_scope("decoder"):
out = input
return (out)