class DDPG(object):
def __init__(self, actor, critic, memory, observation_shape, action_shape, param_noise=None, action_noise=None,
gamma=0.99, tau=0.001, normalize_returns=False, enable_popart=False, normalize_observations=True,
batch_size=128, observation_range=(-5., 5.), action_range=(-1., 1.), return_range=(-np.inf, np.inf),
critic_l2_reg=0., actor_lr=1e-4, critic_lr=1e-3, clip_norm=None, reward_scale=1.):
# Inputs.
self.obs0 = tf.placeholder(tf.float32, shape=(None,) + observation_shape, name='obs0')
self.obs1 = tf.placeholder(tf.float32, shape=(None,) + observation_shape, name='obs1')
self.terminals1 = tf.placeholder(tf.float32, shape=(None, 1), name='terminals1')
self.rewards = tf.placeholder(tf.float32, shape=(None, 1), name='rewards')
self.actions = tf.placeholder(tf.float32, shape=(None,) + action_shape, name='actions')
self.critic_target = tf.placeholder(tf.float32, shape=(None, 1), name='critic_target')
self.param_noise_stddev = tf.placeholder(tf.float32, shape=(), name='param_noise_stddev')
# Parameters.
self.gamma = gamma
self.tau = tau
self.memory = memory
self.normalize_observations = normalize_observations
self.normalize_returns = normalize_returns
self.action_noise = action_noise
self.param_noise = param_noise
self.action_range = action_range
self.return_range = return_range
self.observation_range = observation_range
self.critic = critic
self.actor = actor
self.actor_lr = actor_lr
self.critic_lr = critic_lr
self.clip_norm = clip_norm
self.enable_popart = enable_popart
self.reward_scale = reward_scale
self.batch_size = batch_size
self.stats_sample = None
self.critic_l2_reg = critic_l2_reg
# Observation normalization.
if self.normalize_observations:
with tf.variable_scope('obs_rms'):
self.obs_rms = RunningMeanStd(shape=observation_shape)
else:
self.obs_rms = None
normalized_obs0 = tf.clip_by_value(normalize(self.obs0, self.obs_rms),
self.observation_range[0], self.observation_range[1])
normalized_obs1 = tf.clip_by_value(normalize(self.obs1, self.obs_rms),
self.observation_range[0], self.observation_range[1])
# Return normalization.
if self.normalize_returns:
with tf.variable_scope('ret_rms'):
self.ret_rms = RunningMeanStd()
else:
self.ret_rms = None
# Create target networks.
target_actor = copy(actor)
target_actor.name = 'target_actor'
self.target_actor = target_actor
target_critic = copy(critic)
target_critic.name = 'target_critic'
self.target_critic = target_critic
# Create networks and core TF parts that are shared across setup parts.
self.actor_tf = actor(normalized_obs0)
self.normalized_critic_tf = critic(normalized_obs0, self.actions)
self.critic_tf = denormalize(
tf.clip_by_value(self.normalized_critic_tf, self.return_range[0], self.return_range[1]), self.ret_rms)
self.normalized_critic_with_actor_tf = critic(normalized_obs0, self.actor_tf, reuse=True)
self.critic_with_actor_tf = denormalize(
tf.clip_by_value(self.normalized_critic_with_actor_tf, self.return_range[0], self.return_range[1]),
self.ret_rms)
Q_obs1 = denormalize(target_critic(normalized_obs1, target_actor(normalized_obs1)), self.ret_rms)
self.target_Q = self.rewards + (1. - self.terminals1) * gamma * Q_obs1
# Set up parts.
if self.param_noise is not None:
self.setup_param_noise(normalized_obs0)
self.setup_actor_optimizer()
self.setup_critic_optimizer()
if self.normalize_returns and self.enable_popart:
self.setup_popart()
self.setup_stats()
self.setup_target_network_updates()
def setup_target_network_updates(self):
actor_init_updates, actor_soft_updates = get_target_updates(self.actor.vars, self.target_actor.vars, self.tau)
critic_init_updates, critic_soft_updates = get_target_updates(self.critic.vars, self.target_critic.vars,
self.tau)
self.target_init_updates = [actor_init_updates, critic_init_updates]
self.target_soft_updates = [actor_soft_updates, critic_soft_updates]
def setup_param_noise(self, normalized_obs0):
assert self.param_noise is not None
# Configure perturbed actor.
param_noise_actor = copy(self.actor)
param_noise_actor.name = 'param_noise_actor'
self.perturbed_actor_tf = param_noise_actor(normalized_obs0)
logger.info('setting up param noise')
self.perturb_policy_ops = get_perturbed_actor_updates(self.actor, param_noise_actor, self.param_noise_stddev)
# Configure separate copy for stddev adoption.
adaptive_param_noise_actor = copy(self.actor)
adaptive_param_noise_actor.name = 'adaptive_param_noise_actor'
adaptive_actor_tf = adaptive_param_noise_actor(normalized_obs0)
self.perturb_adaptive_policy_ops = get_perturbed_actor_updates(self.actor, adaptive_param_noise_actor,
self.param_noise_stddev)
self.adaptive_policy_distance = tf.sqrt(tf.reduce_mean(tf.square(self.actor_tf - adaptive_actor_tf)))
def setup_actor_optimizer(self):
logger.info('setting up actor optimizer')
self.actor_loss = -tf.reduce_mean(self.critic_with_actor_tf)
actor_shapes = [var.get_shape().as_list() for var in self.actor.trainable_vars]
actor_nb_params = sum([reduce(lambda x, y: x * y, shape) for shape in actor_shapes])
logger.info(' actor shapes: {}'.format(actor_shapes))
logger.info(' actor params: {}'.format(actor_nb_params))
self.actor_grads = U.flatgrad(self.actor_loss, self.actor.trainable_vars, clip_norm=self.clip_norm)
self.actor_optimizer = Adam(var_list=self.actor.trainable_vars,
beta1=0.9, beta2=0.999, epsilon=1e-08)
def setup_critic_optimizer(self):
logger.info('setting up critic optimizer')
normalized_critic_target_tf = tf.clip_by_value(normalize(self.critic_target, self.ret_rms),
self.return_range[0], self.return_range[1])
self.critic_loss = tf.reduce_mean(tf.square(self.normalized_critic_tf - normalized_critic_target_tf))
if self.critic_l2_reg > 0.:
critic_reg_vars = [var for var in self.critic.trainable_vars if
'kernel' in var.name and 'output' not in var.name]
for var in critic_reg_vars:
logger.info(' regularizing: {}'.format(var.name))
logger.info(' applying l2 regularization with {}'.format(self.critic_l2_reg))
critic_reg = tc.layers.apply_regularization(
tc.layers.l2_regularizer(self.critic_l2_reg),
weights_list=critic_reg_vars
)
self.critic_loss += critic_reg
critic_shapes = [var.get_shape().as_list() for var in self.critic.trainable_vars]
critic_nb_params = sum([reduce(lambda x, y: x * y, shape) for shape in critic_shapes])
logger.info(' critic shapes: {}'.format(critic_shapes))
logger.info(' critic params: {}'.format(critic_nb_params))
self.critic_grads = U.flatgrad(self.critic_loss, self.critic.trainable_vars, clip_norm=self.clip_norm)
self.critic_optimizer = Adam(var_list=self.critic.trainable_vars,
beta1=0.9, beta2=0.999, epsilon=1e-08)
def setup_popart(self):
# See https://arxiv.org/pdf/1602.07714.pdf for details.
self.old_std = tf.placeholder(tf.float32, shape=[1], name='old_std')
new_std = self.ret_rms.std
self.old_mean = tf.placeholder(tf.float32, shape=[1], name='old_mean')
new_mean = self.ret_rms.mean
self.renormalize_Q_outputs_op = []
for vs in [self.critic.output_vars, self.target_critic.output_vars]:
assert len(vs) == 2
M, b = vs
assert 'kernel' in M.name
assert 'bias' in b.name
assert M.get_shape()[-1] == 1
assert b.get_shape()[-1] == 1
self.renormalize_Q_outputs_op += [M.assign(M * self.old_std / new_std)]
self.renormalize_Q_outputs_op += [b.assign((b * self.old_std + self.old_mean - new_mean) / new_std)]
def setup_stats(self):
ops = []
names = []
if self.normalize_returns:
ops += [self.ret_rms.mean, self.ret_rms.std]
names += ['ret_rms_mean', 'ret_rms_std']
if self.normalize_observations:
ops += [tf.reduce_mean(self.obs_rms.mean), tf.reduce_mean(self.obs_rms.std)]
names += ['obs_rms_mean', 'obs_rms_std']
ops += [tf.reduce_mean(self.critic_tf)]
names += ['reference_Q_mean']
ops += [reduce_std(self.critic_tf)]
names += ['reference_Q_std']
ops += [tf.reduce_mean(self.critic_with_actor_tf)]
names += ['reference_actor_Q_mean']
ops += [reduce_std(self.critic_with_actor_tf)]
names += ['reference_actor_Q_std']
ops += [tf.reduce_mean(self.actor_tf)]
names += ['reference_action_mean']
ops += [reduce_std(self.actor_tf)]
names += ['reference_action_std']
if self.param_noise:
ops += [tf.reduce_mean(self.perturbed_actor_tf)]
names += ['reference_perturbed_action_mean']
ops += [reduce_std(self.perturbed_actor_tf)]
names += ['reference_perturbed_action_std']
self.stats_ops = ops
self.stats_names = names
def pi(self, obs, apply_noise=True, compute_Q=True):
if self.param_noise is not None and apply_noise:
actor_tf = self.perturbed_actor_tf
else:
actor_tf = self.actor_tf
feed_dict = {self.obs0: [obs]}
if compute_Q:
action, q = self.sess.run([actor_tf, self.critic_with_actor_tf], feed_dict=feed_dict)
else:
action = self.sess.run(actor_tf, feed_dict=feed_dict)
q = None
action = action.flatten()
if self.action_noise is not None and apply_noise:
noise = self.action_noise()
assert noise.shape == action.shape
action += noise
action = np.clip(action, self.action_range[0], self.action_range[1])
return action, q
def store_transition(self, obs0, action, reward, obs1, terminal1):
reward *= self.reward_scale
self.memory.append(obs0, action, reward, obs1, terminal1)
if self.normalize_observations:
self.obs_rms.update(np.array([obs0]))
def train(self):
# Get a batch.
batch = self.memory.sample(batch_size=self.batch_size)
if self.normalize_returns and self.enable_popart:
old_mean, old_std, target_Q = self.sess.run([self.ret_rms.mean, self.ret_rms.std, self.target_Q],
feed_dict={
self.obs1: batch['obs1'],
self.rewards: batch['rewards'],
self.terminals1: batch['terminals1'].astype('float32'),
})
self.ret_rms.update(target_Q.flatten())
self.sess.run(self.renormalize_Q_outputs_op, feed_dict={
self.old_std: np.array([old_std]),
self.old_mean: np.array([old_mean]),
})
else:
target_Q = self.sess.run(self.target_Q, feed_dict={
self.obs1: batch['obs1'],
self.rewards: batch['rewards'],
self.terminals1: batch['terminals1'].astype('float32'),
})
# Get all gradients and perform a synced update.
ops = [self.actor_grads, self.actor_loss, self.critic_grads, self.critic_loss]
actor_grads, actor_loss, critic_grads, critic_loss = self.sess.run(ops, feed_dict={
self.obs0: batch['obs0'],
self.actions: batch['actions'],
self.critic_target: target_Q,
})
self.actor_optimizer.update(actor_grads, stepsize=self.actor_lr)
self.critic_optimizer.update(critic_grads, stepsize=self.critic_lr)
return critic_loss, actor_loss
def initialize(self, sess):
self.sess = sess
self.sess.run(tf.global_variables_initializer())
self.sess.run(self.target_init_updates)
def update_target_net(self):
self.sess.run(self.target_soft_updates)
def get_stats(self):
if self.stats_sample is None:
# Get a sample and keep that fixed for all further computations.
# This allows us to estimate the change in value for the same set of inputs.
self.stats_sample = self.memory.sample(batch_size=self.batch_size)
values = self.sess.run(self.stats_ops, feed_dict={
self.obs0: self.stats_sample['obs0'],
self.actions: self.stats_sample['actions'],
})
names = self.stats_names[:]
assert len(names) == len(values)
stats = dict(zip(names, values))
if self.param_noise is not None:
stats = {**stats, **self.param_noise.get_stats()}
return stats
def adapt_param_noise(self):
if self.param_noise is None:
return 0.
# Perturb a separate copy of the policy to adjust the scale for the next "real" perturbation.
batch = self.memory.sample(batch_size=self.batch_size)
self.sess.run(self.perturb_adaptive_policy_ops, feed_dict={
self.param_noise_stddev: self.param_noise.current_stddev,
})
distance = self.sess.run(self.adaptive_policy_distance, feed_dict={
self.obs0: batch['obs0'],
self.param_noise_stddev: self.param_noise.current_stddev,
})
self.param_noise.adapt(distance)
return distance
def reset(self):
# Reset internal state after an episode is complete.
if self.action_noise is not None:
self.action_noise.reset()
if self.param_noise is not None:
self.sess.run(self.perturb_policy_ops, feed_dict={
self.param_noise_stddev: self.param_noise.current_stddev,
})
def main():
parser = argparse.ArgumentParser()
parser.add_argument('--D', '-d', type=int, default=8, help='Dimension of feature vector')
parser.add_argument('--T', '-t', type=int, default=2, help='Max step of aggregation')
parser.add_argument('--epoch', '-e', type=int, default=100, help='Number of training dataset')
parser.add_argument('--batch', '-b', type=int, default=256, help='batch size')
parser.add_argument('--flag', '-f', action='store_true', help='make prediction file')
args = parser.parse_args()
train_H, train_y, train_node_size = get_train()
seed = 1996
train_H, train_y, val_H, val_y, train_node_size, val_node_size = shuffle_split(train_H, train_y, train_node_size, split_size=0.7, seed=seed)
# feature dimension
D = args.D
# step size
T = args.T
# learning rate
alpha = 0.015
# epoch size
max_epoch = args.epoch
# batch size
batch_size = args.batch
# get step per epoch
train_size = len(train_H)
iter_per_epoch = train_size//batch_size if (train_size%batch_size) == 0 else (train_size//batch_size)+1
make_pred = args.flag
## make feature vector(train)
train_x = get_feature(D, train_H, train_node_size)
## make feature vector(validation)
val_x = get_feature(D, val_H, val_node_size)
model = GNN(D, T)
optimizer = Adam(alpha=alpha, beta1=0.9, beta2=0.999, eps=1e-8)
train_loss_list = []
train_acc_list = []
val_loss_list = []
val_acc_list = []
for epoch in range(max_epoch):
np.random.seed(int(epoch*1234))
shuffle_idx = np.random.permutation(train_H.shape[0])
train_H = train_H[shuffle_idx]
train_x = train_x[shuffle_idx]
train_y = train_y[shuffle_idx]
for num in range(iter_per_epoch):
if train_size > (num+1)*batch_size:
batch_H = train_H[num*batch_size:(num+1)*batch_size]
batch_x = train_x[num*batch_size:(num+1)*batch_size]
batch_y = train_y[num*batch_size:(num+1)*batch_size]
else:
batch_H = train_H[num*(batch_size):]
batch_x = train_x[num*(batch_size):]
batch_y = train_y[num*(batch_size):]
# get batch gradient and update parameters
batch_grads = None
for idx in range(len(batch_H)):
grad = model.get_gradient(batch_x[idx], batch_H[idx], batch_y[idx])
if batch_grads == None:
batch_grads = {}
for key, val in grad.items():
batch_grads[key] = np.zeros_like(val)
for key in grad.keys():
batch_grads[key] += (grad[key] / len(batch_H))
optimizer.update(model.params, batch_grads)
# train loss and average accuracy
loss = 0
train_pred = np.zeros((len(train_y), 1))
for idx in range(len(train_H)):
loss += model.loss(train_x[idx], train_H[idx], train_y[idx]) / len(train_H)
predict = 0 if model.predict(train_x[idx], train_H[idx]) < 1/2 else 1
train_pred[idx] = predict
train_score = avg_acc(train_y, train_pred)
# validation loss and average accuracy
val_loss = 0
val_pred = np.zeros((len(val_y), 1))
for idx in range(len(val_H)):
val_loss += model.loss(val_x[idx], val_H[idx], val_y[idx]) / len(val_H)
predict = 0 if model.predict(val_x[idx], val_H[idx]) < 1/2 else 1
val_pred[idx] = predict
val_score = avg_acc(val_y, val_pred)
print('epoch:{} loss:{:.5f} val_loss:{:.5f} avg_acc:{:.5f} val_avg_acc:{:.5f}'.format(epoch+1, loss, val_loss, train_score, val_score))
train_loss_list.append(loss)
val_loss_list.append(val_loss)
train_acc_list.append(train_score)
val_acc_list.append(val_score)
fig, (ax1, ax2) = plt.subplots(ncols=2, figsize=(10,4))
x = np.arange(len(train_loss_list))
ax1.plot(x, train_loss_list, label='train')
x = np.arange(len(val_loss_list))
ax1.plot(x, val_loss_list, label='validation')
ax1.legend()
ax1.set_xlabel('epoch')
ax1.set_ylabel('loss')
x = np.arange(len(train_acc_list))
ax2.plot(x, train_acc_list, label='train')
x = np.arange(len(val_acc_list))
ax2.plot(x, val_acc_list, label='validation')
ax2.legend()
ax2.set_xlabel('epoch')
ax2.set_ylabel('average accuracy')
fig.savefig('src/graph/GNN_Adam.png')
plt.close()
if make_pred:
## predict test data
test_H, test_node_size = get_test()
## make feature vector(test)
test_x = get_feature(D, test_H, test_node_size)
with open('prediction.txt', mode='w') as f:
for idx in range(len(test_node_size)):
predict = 0 if model.predict(test_x[idx], test_H[idx]) < 1/2 else 1
f.write('{}'.format(predict) + '\n')
trainer.train()
"""
epoch_num = 0
start = time.time()
for i in range(train_num):
batch_mask = np.random.choice(x_train.shape[0], batch_size)
x_batch = x_train[batch_mask]
t_batch = t_train[batch_mask]
batch_mask_test = np.random.choice(x_test.shape[0], batch_size)
x_test_batch = x_test[batch_mask_test]
t_test_batch = t_test[batch_mask_test]
grad = network.gradient(x_batch, t_batch)
# パラメータの更新
optimizer.update(network.params, grad)
#for key in grad.keys():
# network.params[key] -= learning_rate * grad[key]
# 損失関数計算
loss = network.loss(x_batch, t_batch)
train_loss_list.append(loss)
train_acc = network.accuracy(x_batch, t_batch)
test_acc = network.accuracy(x_test_batch, t_test_batch)
train_acc_list.append(train_acc)
test_acc_list.append(test_acc)
print(".", end="", flush=True)
