深層強化学習 CartPole-v0 を動かしてみる（PyTorch のサンプルプログラムを使用）

ユースケース: 深層強化学習 CartPole-v0 を動かしてみたい

PyTorch の Web ページ: http://pytorch.org

GitHub の PyTorch の Webページ: https://github.com/pytorch/pytorch

前準備

Open-AI gym のインストール

• Ubuntu の場合

端末で，次のコマンドを実行．

sudo pip3 install gym 深層強化学習 CartPole-v0 を動かしてみる（PyTorch のサンプルプログラムを使用）

Python プログラムを動かす．

Python プログラムを動かすために， Windows では「python」， Ubuntu では「python3」などのコマンドを使う．

あるいは， 開発環境や Python コンソール（Jupyter Qt ConsoleSpyderPyCharmPyScripter など）の利用も便利である．

1. インポート

import gym
import math
import random
import numpy as np
import matplotlib
%matplotlib inline
import matplotlib.pyplot as plt
import warnings
warnings.filterwarnings('ignore')   # Suppress Matplotlib warnings
from collections import namedtuple
from itertools import count
from PIL import Image

import torch
import torch.nn as nn
import torch.optim as optim
import torch.nn.functional as F
import torchvision.transforms as T

env = gym.make('CartPole-v0').unwrapped

# set up matplotlib
is_ipython = 'inline' in matplotlib.get_backend()
if is_ipython:
from IPython import display

plt.ion()

# if gpu is to be used
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")

Ubuntu での実行結果例 2. CIFAR 10 のダウンロード

transform = transforms.Compose(
[transforms.ToTensor(),
transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))])

classes = ('plane', 'car', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck')

Ubuntu の PyCharm での実行結果例 3. Replay Memory の作成

Transition = namedtuple('Transition',
('state', 'action', 'next_state', 'reward'))

class ReplayMemory(object):

def __init__(self, capacity):
self.capacity = capacity
self.memory = []
self.position = 0

def push(self, *args):
"""Saves a transition."""
if len(self.memory) < self.capacity:
self.memory.append(None)
self.memory[self.position] = Transition(*args)
self.position = (self.position + 1) % self.capacity

def sample(self, batch_size):
return random.sample(self.memory, batch_size)

def __len__(self):
return len(self.memory) 4. Q-network の作成

class DQN(nn.Module):

def __init__(self, h, w, outputs):
super(DQN, self).__init__()
self.conv1 = nn.Conv2d(3, 16, kernel_size=5, stride=2)
self.bn1 = nn.BatchNorm2d(16)
self.conv2 = nn.Conv2d(16, 32, kernel_size=5, stride=2)
self.bn2 = nn.BatchNorm2d(32)
self.conv3 = nn.Conv2d(32, 32, kernel_size=5, stride=2)
self.bn3 = nn.BatchNorm2d(32)

# Number of Linear input connections depends on output of conv2d layers
# and therefore the input image size, so compute it.
def conv2d_size_out(size, kernel_size = 5, stride = 2):
return (size - (kernel_size - 1) - 1) // stride  + 1
convw = conv2d_size_out(conv2d_size_out(conv2d_size_out(w)))
convh = conv2d_size_out(conv2d_size_out(conv2d_size_out(h)))
linear_input_size = convw * convh * 32

# Called with either one element to determine next action, or a batch
# during optimization. Returns tensor([[left0exp,right0exp]...]).
def forward(self, x):
x = F.relu(self.bn1(self.conv1(x)))
x = F.relu(self.bn2(self.conv2(x)))
x = F.relu(self.bn3(self.conv3(x))) 5. Input Extraction (入力の抽出)

resize = T.Compose([T.ToPILImage(),
T.Resize(40, interpolation=Image.CUBIC),
T.ToTensor()])

def get_cart_location(screen_width):
world_width = env.x_threshold * 2
scale = screen_width / world_width
return int(env.state * scale + screen_width / 2.0)  # MIDDLE OF CART

def get_screen():
# Returned screen requested by gym is 400x600x3, but is sometimes larger
# such as 800x1200x3. Transpose it into torch order (CHW).
screen = env.render(mode='rgb_array').transpose((2, 0, 1))
# Cart is in the lower half, so strip off the top and bottom of the screen
_, screen_height, screen_width = screen.shape
screen = screen[:, int(screen_height*0.4):int(screen_height * 0.8)]
view_width = int(screen_width * 0.6)
cart_location = get_cart_location(screen_width)
if cart_location < view_width // 2:
slice_range = slice(view_width)
elif cart_location > (screen_width - view_width // 2):
slice_range = slice(-view_width, None)
else:
slice_range = slice(cart_location - view_width // 2,
cart_location + view_width // 2)
# Strip off the edges, so that we have a square image centered on a cart
screen = screen[:, :, slice_range]
# Convert to float, rescale, convert to torch tensor
# (this doesn't require a copy)
screen = np.ascontiguousarray(screen, dtype=np.float32) / 255
screen = torch.from_numpy(screen)
# Resize, and add a batch dimension (BCHW)
return resize(screen).unsqueeze(0).to(device)

env.reset()
plt.figure()
plt.imshow(get_screen().cpu().squeeze(0).permute(1, 2, 0).numpy(),
interpolation='none')
plt.title('Example extracted screen')
plt.show() 6. ハイパーパラメータなど

BATCH_SIZE = 128
GAMMA = 0.999
EPS_START = 0.9
EPS_END = 0.05
EPS_DECAY = 200
TARGET_UPDATE = 10

# Get screen size so that we can initialize layers correctly based on shape
# returned from AI gym. Typical dimensions at this point are close to 3x40x90
# which is the result of a clamped and down-scaled render buffer in get_screen()
init_screen = get_screen()
_, _, screen_height, screen_width = init_screen.shape

# Get number of actions from gym action space
n_actions = env.action_space.n

policy_net = DQN(screen_height, screen_width, n_actions).to(device)
target_net = DQN(screen_height, screen_width, n_actions).to(device)
target_net.eval()

optimizer = optim.RMSprop(policy_net.parameters())
memory = ReplayMemory(10000)

steps_done = 0

def select_action(state):
global steps_done
sample = random.random()
eps_threshold = EPS_END + (EPS_START - EPS_END) * \
math.exp(-1. * steps_done / EPS_DECAY)
steps_done += 1
if sample > eps_threshold:
# t.max(1) will return largest column value of each row.
# second column on max result is index of where max element was
# found, so we pick action with the larger expected reward.
return policy_net(state).max(1).view(1, 1)
else:

episode_durations = []

def plot_durations():
plt.figure(2)
plt.clf()
durations_t = torch.tensor(episode_durations, dtype=torch.float)
plt.title('Training...')
plt.xlabel('Episode')
plt.ylabel('Duration')
plt.plot(durations_t.numpy())
# Take 100 episode averages and plot them too
if len(durations_t) >= 100:
means = durations_t.unfold(0, 100, 1).mean(1).view(-1)
means = torch.cat((torch.zeros(99), means))
plt.plot(means.numpy())

plt.pause(0.001)  # pause a bit so that plots are updated
if is_ipython:
display.clear_output(wait=True)
display.display(plt.gcf()) 7. 最適化の1ステップを実行する関数

def optimize_model():
if len(memory) < BATCH_SIZE:
return
transitions = memory.sample(BATCH_SIZE)
# Transpose the batch (see https://stackoverflow.com/a/19343/3343043 for
# detailed explanation). This converts batch-array of Transitions
# to Transition of batch-arrays.
batch = Transition(*zip(*transitions))

# Compute a mask of non-final states and concatenate the batch elements
# (a final state would've been the one after which simulation ended)
non_final_mask = torch.tensor(tuple(map(lambda s: s is not None,
batch.next_state)), device=device, dtype=torch.uint8)
non_final_next_states = torch.cat([s for s in batch.next_state
if s is not None])
state_batch = torch.cat(batch.state)
action_batch = torch.cat(batch.action)
reward_batch = torch.cat(batch.reward)

# Compute Q(s_t, a) - the model computes Q(s_t), then we select the
# columns of actions taken. These are the actions which would've been taken
# for each batch state according to policy_net
state_action_values = policy_net(state_batch).gather(1, action_batch)

# Compute V(s_{t+1}) for all next states.
# Expected values of actions for non_final_next_states are computed based
# on the "older" target_net; selecting their best reward with max(1).
# This is merged based on the mask, such that we'll have either the expected
# state value or 0 in case the state was final.
next_state_values = torch.zeros(BATCH_SIZE, device=device)
# Compute the expected Q values
expected_state_action_values = (next_state_values * GAMMA) + reward_batch

# Compute Huber loss
loss = F.smooth_l1_loss(state_action_values, expected_state_action_values.unsqueeze(1))

# Optimize the model
loss.backward()
for param in policy_net.parameters():
optimizer.step() 8. 学習

num_episodes = 50
for i_episode in range(num_episodes):
# Initialize the environment and state
env.reset()
last_screen = get_screen()
current_screen = get_screen()
state = current_screen - last_screen
for t in count():
# Select and perform an action
action = select_action(state)
_, reward, done, _ = env.step(action.item())
reward = torch.tensor([reward], device=device)

# Observe new state
last_screen = current_screen
current_screen = get_screen()
if not done:
next_state = current_screen - last_screen
else:
next_state = None

# Store the transition in memory
memory.push(state, action, next_state, reward)

# Move to the next state
state = next_state

# Perform one step of the optimization (on the target network)
optimize_model()
if done:
episode_durations.append(t + 1)
plot_durations()
break
# Update the target network, copying all weights and biases in DQN
if i_episode % TARGET_UPDATE == 0: 