我正在尝试将get_gradients(loss, params)函数从this file移植到keras-rl中的DDPG.py代理，但我在执行方面遇到了一些问题。
我修改了DDPG.py中的action_bounds = [-30, 50]
inverted_grads = []
for g,p in zip(modified_grads, params):
is_above_upper_bound = K.greater(p, K.constant(action_bounds[1], dtype='float32'))
is_under_lower_bound = K.less(p, K.constant(action_bounds[0], dtype='float32'))
is_gradient_positive = K.greater(g, K.constant(0, dtype='float32'))
is_gradient_negative = K.less(g, K.constant(0, dtype='float32'))
invert_gradient = tf.logical_or(
tf.logical_and(is_above_upper_bound, is_gradient_negative),
tf.logical_and(is_under_lower_bound, is_gradient_positive)
)
inverted_grads.extend(K.switch(invert_gradient, -g, g))
modified_grads = inverted_grads[:]
方法以添加此内容：
ValueError: Shape must be rank 0 but is rank 2 for 'cond/Switch' (op: 'Switch') with input shapes: [2,400], [2,400].
但我收到有关形状的错误：
var data = ko.toJSON(self.Profiles());
$.ajax({
type: 'POST',
url: '/ajaxCall/insertProAjax',
data: "{ Customer:" + ko.utils.stringifyJson(self.Name) + ",customerRemove:" + ko.utils.stringifyJson(self.CustomerRemove) + "}",
contentType: "application/json",
success: function (data) {
alert("Record has been saved Successfully");
MarkCustomerAsSaved();
$('#AddNewModel').modal('hide');
},
error: function () {
alert("Failed");
}
}).fail(function (xhr, textStatus, err) { alert(err); });
答案 0 :(得分：1)
keras-rl“get_gradients”函数使用通过组合的actor-critic模型计算的渐变，但是您需要使用动作输入的批评输出的渐变来应用反转渐变特征。
我最近使用keras-rl在我正在研究的RDPG原型上实现了它。仍在测试，代码可以被优化，并且肯定没有bug，但我通过修改一些keras-rl代码行来使反转渐变起作用。为了修改动作输入的评论输出的渐变，我已经按照原始公式来计算演员渐变，借助Patrick Emami的这篇精彩文章：http://pemami4911.github.io/blog/2016/08/21/ddpg-rl.html。
我在这里放置整个“编译”函数，在继承自“DDPAgent”的类中重新定义，其中实现了反转渐变功能。
def compile(self, optimizer, metrics=[]):
metrics += [mean_q]
if type(optimizer) in (list, tuple):
if len(optimizer) != 2:
raise ValueError('More than two optimizers provided. Please only provide a maximum of two optimizers, the first one for the actor and the second one for the critic.')
actor_optimizer, critic_optimizer = optimizer
else:
actor_optimizer = optimizer
critic_optimizer = clone_optimizer(optimizer)
if type(actor_optimizer) is str:
actor_optimizer = optimizers.get(actor_optimizer)
if type(critic_optimizer) is str:
critic_optimizer = optimizers.get(critic_optimizer)
assert actor_optimizer != critic_optimizer
if len(metrics) == 2 and hasattr(metrics[0], '__len__') and hasattr(metrics[1], '__len__'):
actor_metrics, critic_metrics = metrics
else:
actor_metrics = critic_metrics = metrics
def clipped_error(y_true, y_pred):
return K.mean(huber_loss(y_true, y_pred, self.delta_clip), axis=-1)
# Compile target networks. We only use them in feed-forward mode, hence we can pass any
# optimizer and loss since we never use it anyway.
self.target_actor = clone_model(self.actor, self.custom_model_objects)
self.target_actor.compile(optimizer='sgd', loss='mse')
self.target_critic = clone_model(self.critic, self.custom_model_objects)
self.target_critic.compile(optimizer='sgd', loss='mse')
# We also compile the actor. We never optimize the actor using Keras but instead compute
# the policy gradient ourselves. However, we need the actor in feed-forward mode, hence
# we also compile it with any optimzer and
self.actor.compile(optimizer='sgd', loss='mse')
# Compile the critic.
if self.target_model_update < 1.:
# We use the `AdditionalUpdatesOptimizer` to efficiently soft-update the target model.
critic_updates = get_soft_target_model_updates(self.target_critic, self.critic, self.target_model_update)
critic_optimizer = AdditionalUpdatesOptimizer(critic_optimizer, critic_updates)
self.critic.compile(optimizer=critic_optimizer, loss=clipped_error, metrics=critic_metrics)
clipnorm = getattr(actor_optimizer, 'clipnorm', 0.)
clipvalue = getattr(actor_optimizer, 'clipvalue', 0.)
critic_gradients_wrt_action_input = tf.gradients(self.critic.output, self.critic_action_input)
critic_gradients_wrt_action_input = [g / float(self.batch_size) for g in critic_gradients_wrt_action_input] # since TF sums over the batch
action_bounds = [(-1.,1.) for i in range(self.nb_actions)]
def calculate_inverted_gradient():
"""
Applies "inverting gradient" feature to the action-value gradients.
"""
gradient_wrt_action = -critic_gradients_wrt_action_input[0]
inverted_gradients = []
for n in range(self.batch_size):
inverted_gradient = []
for i in range(gradient_wrt_action[n].shape[0].value):
action = self.critic_action_input[n][i]
is_gradient_negative = K.less(gradient_wrt_action[n][i], K.constant(0, dtype='float32'))
adjust_for_upper_bound = gradient_wrt_action[n][i] * ((action_bounds[i][1] - action) / (action_bounds[i][1] - action_bounds[i][0]))
adjust_for_lower_bound = gradient_wrt_action[n][i] * ((action - action_bounds[i][0]) / (action_bounds[i][1] - action_bounds[i][0]))
modified_gradient = K.switch(is_gradient_negative, adjust_for_upper_bound, adjust_for_lower_bound)
inverted_gradient.append( modified_gradient )
inverted_gradients.append(inverted_gradient)
gradient_wrt_action = tf.stack(inverted_gradients)
return gradient_wrt_action
actor_gradients_wrt_weights = tf.gradients(self.actor.output, self.actor.trainable_weights, grad_ys=calculate_inverted_gradient())
actor_gradients_wrt_weights = [g / float(self.batch_size) for g in actor_gradients_wrt_weights] # since TF sums over the batch
def get_gradients(loss, params):
""" Used by the actor optimizer.
Returns the gradients to train the actor.
These gradients are obtained by multiplying the gradients of the actor output w.r.t. its weights
with the gradients of the critic output w.r.t. its action input. """
# Aplly clipping if defined
modified_grads = [g for g in actor_gradients_wrt_weights]
if clipnorm > 0.:
norm = K.sqrt(sum([K.sum(K.square(g)) for g in modified_grads]))
modified_grads = [optimizers.clip_norm(g, clipnorm, norm) for g in modified_grads]
if clipvalue > 0.:
modified_grads = [K.clip(g, -clipvalue, clipvalue) for g in modified_grads]
return modified_grads
actor_optimizer.get_gradients = get_gradients
# get_updates is the optimizer function that changes the weights of the network
updates = actor_optimizer.get_updates(self.actor.trainable_weights, self.actor.constraints, None)
if self.target_model_update < 1.:
# Include soft target model updates.
updates += get_soft_target_model_updates(self.target_actor, self.actor, self.target_model_update)
updates += self.actor.updates # include other updates of the actor, e.g. for BN
# Finally, combine it all into a callable function.
# The inputs will be all the necessary placeholders to compute the gradients (actor and critic inputs)
inputs = self.actor.inputs[:] + [self.critic_action_input, self.critic_history_input]
self.actor_train_fn = K.function(inputs, [self.actor.output], updates=updates)
self.actor_optimizer = actor_optimizer
self.compiled = True
训练演员时，你现在应该传递3个输入而不是2：观察输入+动作输入（使用演员网络的预测），所以你还必须修改“向后”功能。就我而言：
...
if self.episode > self.nb_steps_warmup_actor:
action = self.actor.predict_on_batch(history_batch)
inputs = [history_batch, action, history_batch]
actor_train_result = self.actor_train_fn(inputs)
action_values = actor_train_result[0]
assert action_values.shape == (self.batch_size, self.nb_actions)
...
之后，你可以让你的演员在输出中进行线性激活。