EPOCH = 2000
LR = 1e-5
cwd = '/home/qinlong/PycharmProjects/NEU/w_b_transfer/w_data/'
# input
x = tf.placeholder(tf.float32, shape=[BATCH_SIZE,5], name='structure_parameter')
para_train, id_train, sp_train = tfrecord.read_tfrecord(cwd + '/train_data.tfrecord', BATCH_SIZE)
para_test, id_test, sp_test = tfrecord.read_tfrecord(cwd + '/test_data.tfrecord', BATCH_SIZE)
test_dataset = tfrecord.read_tfrecord(cwd + 'train_data.tfrecord', BATCH_SIZE)
# loss
r_spec = tf.placeholder(tf.float32, shape=[BATCH_SIZE, 603], name='r_spec')
p_spec = fn.fw_net(x)
loss = ms.huber_loss(r_spec, p_spec)
# optimizer
optimizer = tf.train.AdamOptimizer(learning_rate=LR,
beta1=0.9,
beta2=0.999,
epsilon=1e-8)
# minimize
fw_op = optimizer.minimize(loss=loss)
# start training
init = tf.global_variables_initializer()
train_summary = tf.summary.scalar('Train loss', loss)
test_summary = tf.summary.scalar('Test loss', loss)
t = ms.conv(t, 32, [3, 3])
# x = ms.bn(x, training=training)
t = ms.activation(t, relu=True)
t = ms.conv(t, 32, [3, 3])
# x = ms.bn(x, training=training)
t = ms.activation(t, relu=True)
t = tf.reshape(t, [-1, 40 * 40 * 32])
with tf.variable_scope('fc2'):
t = ms.fc(t, units=402)
# x = ms.bn(x, training=training)
t = ms.activation(t, relu=False)
new_loss = ms.huber_loss(t, y_)
new_optimizer = tf.train.AdamOptimizer(learning_rate=LR,
beta1=0.9,
beta2=0.999,
epsilon=1e-8,
name='new_adam')
new_op = new_optimizer.minimize(loss=new_loss)
var = tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES,
scope='trans_part')
sess.run(tf.variables_initializer(var_list=var))
sess.run(tf.variables_initializer(new_optimizer.variables()))
# summary tensorboard