本文接上一篇博文：
一、程序代码
程序主要实现上篇文章中所提到的随机噪声拟合高斯分布的过程，话不多说，直接上代码：
#引入必要的包
import argparse
import numpy as np
from scipy.stats import norm
import tensorflow as tf
import matplotlib.pyplot as plt
import seaborn as sns
sns.set(color_codes=True)
#设置种子，用于随机初始化
seed = 42
np.random.seed(seed)
tf.set_random_seed(seed)
#定义真实的数据分布，这里为高斯分布
class DataDistribution(object):
def __init__(self):
#高斯分布参数
#均值为4
self.mu = 4
#标准差为0.5
self.sigma = 0.5
def sample(self, N):
samples = np.random.normal(self.mu, self.sigma, N)
samples.sort()
return samples
#随机初始化一个分布，做为G网络的输入
class GeneratorDistribution(object):
def __init__(self, range):
self.range = range
def sample(self, N):
return np.linspace(-self.range, self.range, N) + \
np.random.random(N) * 0.01
#定义线性运算函数，其中参数output_dim=h_dim*2=8
def linear(input, output_dim, scope=None, stddev=1.0):
#定义一个随机初始化
norm = tf.random_normal_initializer(stddev=stddev)
#b初始化为0
const = tf.constant_initializer(0.0)
with tf.variable_scope(scope or 'linear'):
#声明w的shape,输入为（12，1）*w，故w为（1，8），w的初始化方式为高斯初始化
w = tf.get_variable('w', [input.get_shape()[1], output_dim], initializer=norm)
#b初始化为常量
b = tf.get_variable('b', [output_dim], initializer=const)
#执行线性运算
return tf.matmul(input, w) + b
#
def generator(input, h_dim):
h0 = tf.nn.softplus(linear(input, h_dim, 'g0'))
h1 = linear(h0, 1, 'g1')
return h1
#初始化w和b的函数，其中h0,h1,h2,h3为层，将mlp_hidden_size=4传给h_dim
def discriminator(input, h_dim):
#linear 控制w和b的初始化,这里linear函数的第二个参数为4*2=8
#第一层
h0 = tf.tanh(linear(input, h_dim * 2, 'd0'))
#第二层输出，隐藏层神经元个数还是为8
h1 = tf.tanh(linear(h0, h_dim * 2, 'd1'))
#h2为第三层输出值
h2 = tf.tanh(linear(h1, h_dim * 2, scope='d2'))
#最终的输出值
h3 = tf.sigmoid(linear(h2, 1, scope='d3'))
return h3
#优化器 采用学习率衰减的方法
def optimizer(loss, var_list, initial_learning_rate):
decay = 0.95
num_decay_steps = 150
batch = tf.Variable(0)
#调用学习率衰减的函数
learning_rate = tf.train.exponential_decay(
initial_learning_rate,
batch,
num_decay_steps,
decay,
staircase=True
)
#梯度下降求解
optimizer = tf.train.GradientDescentOptimizer(learning_rate).minimize(
loss,
global_step=batch,
var_list=var_list
)
#返回
return optimizer
#构造模型
class GAN(object):
def __init__(self, data, gen, num_steps, batch_size, log_every):
self.data = data
self.gen = gen
self.num_steps = num_steps
self.batch_size = batch_size
self.log_every = log_every
#隐藏层神经元个数
self.mlp_hidden_size = 4
#学习率
self.learning_rate = 0.03
#通过placeholder格式来创造模型
self._create_model()
def _create_model(self):
#创建一个名叫D_pre的域，先构造一个D_pre网络，用来训练出真正D网络初始化网络所需要的参数
with tf.variable_scope('D_pre'):
#输入的shape为（12，1），一个batch一个batch的训练，
#每个batch的大小为12，要训练的数据为1维的点
self.pre_input = tf.placeholder(tf.float32, shape=(self.batch_size, 1))
self.pre_labels = tf.placeholder(tf.float32, shape=(self.batch_size, 1))
#调用discriminator来初始化w和b参数,其中self.mlp_hidden_size=4，为discriminator函数的第二个参数
D_pre = discriminator(self.pre_input, self.mlp_hidden_size)
#预测值和label之间的差异
self.pre_loss = tf.reduce_mean(tf.square(D_pre - self.pre_labels))
#定义优化器求解
self.pre_opt = optimizer(self.pre_loss, None, self.learning_rate)
# This defines the generator network - it takes samples from a noise
# distribution as input, and passes them through an MLP.
#真正的G网络
with tf.variable_scope('Gen'):
self.z = tf.placeholder(tf.float32, shape=(self.batch_size, 1))
#生产网络只有两层
self.G = generator(self.z, self.mlp_hidden_size)
# The discriminator tries to tell the difference between samples from the
# true data distribution (self.x) and the generated samples (self.z).
#
# Here we create two copies of the discriminator network (that share parameters),
# as you cannot use the same network with different inputs in TensorFlow.
#D网络
with tf.variable_scope('Disc') as scope:
self.x = tf.placeholder(tf.float32, shape=(self.batch_size, 1))
#构造D1网络，真实的数据
self.D1 = discriminator(self.x, self.mlp_hidden_size)
#重新使用一下变量，不用重新定义
scope.reuse_variables()
#D2，生成的数据
self.D2 = discriminator(self.G, self.mlp_hidden_size)
# Define the loss for discriminator and generator networks (see the original
# paper for details), and create optimizers for both
#定义判别网络损失函数
self.loss_d = tf.reduce_mean(-tf.log(self.D1) - tf.log(1 - self.D2))
#定义生成网络损失函数
self.loss_g = tf.reduce_mean(-tf.log(self.D2))
self.d_pre_params = tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES, scope='D_pre')
self.d_params = tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES, scope='Disc')
self.g_params = tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES, scope='Gan')
#优化，得到两组参数
self.opt_d = optimizer(self.loss_d, self.d_params, self.learning_rate)
self.opt_g = optimizer(self.loss_g, self.g_params, self.learning_rate)
def train(self):
with tf.Session() as session:
tf.global_variables_initializer().run()
# pretraining discriminator
#先训练D_pre网络
num_pretrain_steps = 1000
for step in range(num_pretrain_steps):
#随机生成数据
d = (np.random.random(self.batch_size) - 0.5) * 10.0
labels = norm.pdf(d, loc=self.data.mu, scale=self.data.sigma)
pretrain_loss, _ = session.run([self.pre_loss, self.pre_opt], {
self.pre_input: np.reshape(d, (self.batch_size, 1)),
self.pre_labels: np.reshape(labels, (self.batch_size, 1))
})
#拿出预训练好的数据
self.weightsD = session.run(self.d_pre_params)
# copy weights from pre-training over to new D network
for i, v in enumerate(self.d_params):
session.run(v.assign(self.weightsD[i]))
#训练真正的网络
for step in range(self.num_steps):
# update discriminator
x = self.data.sample(self.batch_size)
#z是一个随机生成的噪音
z = self.gen.sample(self.batch_size)
#优化判别网络
loss_d, _ = session.run([self.loss_d, self.opt_d], {
self.x: np.reshape(x, (self.batch_size, 1)),
self.z: np.reshape(z, (self.batch_size, 1))
})
# update generator
#随机初始化
z = self.gen.sample(self.batch_size)
#迭代优化
loss_g, _ = session.run([self.loss_g, self.opt_g], {
self.z: np.reshape(z, (self.batch_size, 1))
})
#打印
if step % self.log_every == 0:
print('{}: {}\t{}'.format(step, loss_d, loss_g))
#画图
if step % 100 == 0 or step==0 or step == self.num_steps -1 :
self._plot_distributions(session)
def _samples(self, session, num_points=10000, num_bins=100):
xs = np.linspace(-self.gen.range, self.gen.range, num_points)
bins = np.linspace(-self.gen.range, self.gen.range, num_bins)
# data distribution
d = self.data.sample(num_points)
pd, _ = np.histogram(d, bins=bins, density=True)
# generated samples
zs = np.linspace(-self.gen.range, self.gen.range, num_points)
g = np.zeros((num_points, 1))
for i in range(num_points // self.batch_size):
g[self.batch_size * i:self.batch_size * (i + 1)] = session.run(self.G, {
self.z: np.reshape(
zs[self.batch_size * i:self.batch_size * (i + 1)],
(self.batch_size, 1)
)
})
pg, _ = np.histogram(g, bins=bins, density=True)
return pd, pg
def _plot_distributions(self, session):
pd, pg = self._samples(session)
p_x = np.linspace(-self.gen.range, self.gen.range, len(pd))
f, ax = plt.subplots(1)
ax.set_ylim(0, 1)
plt.plot(p_x, pd, label='real data')
plt.plot(p_x, pg, label='generated data')
plt.title('1D Generative Adversarial Network')
plt.xlabel('Data values')
plt.ylabel('Probability density')
plt.legend()
plt.show()
def main(args):
model = GAN(
#定义真实数据的分布
DataDistribution(),
#创造一些噪音点，用来传入G函数
GeneratorDistribution(range=8),
#迭代次数
args.num_steps,
#一次迭代12个点的数据
args.batch_size,
#隔多少次打印当前loss
args.log_every,
)
model.train()
def parse_args():
parser = argparse.ArgumentParser()
parser.add_argument('--num-steps', type=int, default=3000,
help='the number of training steps to take')
parser.add_argument('--batch-size', type=int, default=12,
help='the batch size')
parser.add_argument('--log-every', type=int, default=10,
help='print loss after this many steps')
return parser.parse_args()
if __name__ == '__main__':
main(parse_args())二、程序运行结果
1、程序运行初始状态
其中左边为随机初始化的数据，右边为真实的呈高斯分布的数据。
2、程序迭代运行1200次后的状态
这里不知道为什么原因，程序没有正常的拟合真实的数据，将迭代次数增加之后，程序也没有太大的变化，D和G网络的两个Loss的变化都很小，这里还望大家帮忙找一找原因。可能和GAN网络容易训练跑偏的一些原因有关。