tensorflow40《TensorFlow实战》笔记-08-01 TensorFlow实现深度强化学习-策略网络 code

# 《TensorFlow实战》08 TensorFlow实现深度强化学习
# win10 Tensorflow1.0.1 python3.5.3
# CUDA v8.0 cudnn-8.0-windows10-x64-v5.1
# filename:sz08.01.py # TensorFlow实现策略网络

# https://github.com/awjuliani/DeepRL-Agents.git
# .ipybn文件使用jupyter notebook打开
# https://github.com/awjuliani/DeepRL-Agents/blob/master/Double-Dueling-DQN.ipynb
# pip3 install gym

import numpy as np
import tensorflow as tf
import gym
env = gym.make('CartPole-v0')
env.reset()
random_episodes = 0
reward_sum = 0
while random_episodes < 10:
    env.render()
    observation, reward, done, _ = env.step(np.random.randint(0,2))
    reward_sum += reward
    if done:
        random_episodes += 1
        print('Reward for this episode was:', reward_sum)
        reward_sum = 0
        env.reset()
'''
Reward for this episode was: 52.0
Reward for this episode was: 22.0
Reward for this episode was: 15.0
Reward for this episode was: 38.0
Reward for this episode was: 19.0
Reward for this episode was: 8.0
Reward for this episode was: 19.0
Reward for this episode was: 18.0
Reward for this episode was: 34.0
Reward for this episode was: 10.0
'''

H = 50
batch_size = 25
learning_rate = 1e-1
D = 4
gamma = 0.99

observations = tf.placeholder(tf.float32, [None, D], name='input_x')
W1 = tf.get_variable('W1', shape=[D, H], initializer=tf.contrib.layers.xavier_initializer())
layer1 = tf.nn.relu(tf.matmul(observations, W1))
W2 = tf.get_variable('W2', shape=[H, 1], initializer=tf.contrib.layers.xavier_initializer())
score = tf.matmul(layer1, W2)
probability = tf.nn.sigmoid(score)
adam = tf.train.AdamOptimizer(learning_rate=learning_rate)
W1Grad = tf.placeholder(tf.float32, name='batch_grad1')
W2Grad = tf.placeholder(tf.float32, name='batch_grad2')
batchGrad = [W1Grad, W2Grad]

def discount_rewards(r):
    discounted_r = np.zeros_like(r)
    running_add = 0
    for t in reversed(range(r.size)):
        running_add = running_add * gamma + r[t]
        discounted_r[t] = running_add
    return discounted_r

input_y = tf.placeholder(tf.float32, [None, 1], name='input_y')
advantages = tf.placeholder(tf.float32, name='reward_signal')
loglik = tf.log(input_y*(input_y - probability) + (1 - input_y)*(input_y + probability))
loss = -tf.reduce_mean(loglik* advantages)
tvars = tf.trainable_variables()
newGrads = tf.gradients(loss, tvars)
updateGrads = adam.apply_gradients(zip(batchGrad, tvars))

xs, ys, drs = [], [], []
reward_sum = 0
episode_number = 1
total_episodes = 400
with tf.Session() as sess:
    rendering = False
    init = tf.global_variables_initializer()
    sess.run(init)
    observation = env.reset()
    gradBuffer = sess.run(tvars)
    for ix, grad in enumerate(gradBuffer):
        gradBuffer[ix] = grad * 0

    while episode_number <= total_episodes:
        if reward_sum/batch_size > 100 or rendering == True:
            env.render()
            rendering = True

        x = np.reshape(observation, [1, D])

        tfprob = sess.run(probability, feed_dict={observations: x})
        action = 1 if np.random.uniform() < tfprob else 0

        xs.append(x)
        y = 1 - action
        ys.append(y)

        observation, reward, done, info = env.step(action)
        reward_sum += reward
        drs.append(reward)
        if done:
            episode_number += 1
            epx = np.vstack(xs)
            epy = np.vstack(ys)
            epr = np.vstack(drs)
            xs, ys, drs = [],[],[]

            discounted_epr = discount_rewards(epr)
            discounted_epr -= np.mean(discounted_epr)
            discounted_epr /= np.std(discounted_epr)

            tGrad = sess.run(newGrads, feed_dict={observations: epx, input_y: epy, advantages:discounted_epr})
            for ix, grad in enumerate(tGrad):
                gradBuffer[ix] += grad

            if episode_number % batch_size == 0:
                sess.run(updateGrads, feed_dict={W1Grad: gradBuffer[0],W2Grad:gradBuffer[1]})
                for ix, grad in enumerate(gradBuffer):
                    gradBuffer[ix] = grad * 0
                print('Average reward for episode %d: %f.' % (episode_number, reward_sum/batch_size))
                if reward_sum/batch_size > 200:
                    print("Task solved in ", episode_number, 'episodes!')
                    break
                reward_sum = 0
            observation = env.reset()
'''
Average reward for episode 25: 21.200000.
Average reward for episode 50: 37.240000.
Average reward for episode 75: 59.600000.
Average reward for episode 100: 76.200000.
Average reward for episode 125: 80.280000.
Average reward for episode 150: 88.360000.
Average reward for episode 175: 98.960000.
Average reward for episode 200: 142.120000.
Average reward for episode 225: 179.840000.
Average reward for episode 250: 193.760000.
Average reward for episode 275: 199.600000.
Average reward for episode 300: 198.240000.
Average reward for episode 325: 200.000000.
Average reward for episode 350: 200.000000.
Average reward for episode 375: 200.000000.
Average reward for episode 400: 200.000000.
'''