Scale actor network output to the action space bounds in Keras Rl
0
I am trying to implement DDPG from Keras RL and have the following actor network.
actor = Sequential()
actor.add(Flatten(input_shape=(1,) + env.observation_space.shape))
actor.add(Dense(16))
actor.add(Activation('relu'))
actor.add(Dense(16))
actor.add(Activation('relu'))
actor.add(Dense(16))
actor.add(Activation('relu'))
actor.add(Dense(nb_actions))
actor.add(Activation('linear'))
However, I would prefer to have the output scaled to a custom gym environment action space bounds for my problem. env.action_space.
https://pemami4911.github.io/blog/2016/08/21/ddpg-rl.html shows this using the tflearn api where they use
def create_actor_network(self):
inputs = tflearn.input_data(shape=[None, self.s_dim])
net = tflearn.fully_connected(inputs, 400)
net = tflearn.layers.normalization.batch_normalization(net)
net = tflearn.activations.relu(net)
net = tflearn.fully_connected(net, 300)
net = tflearn.layers.normalization.batch_normalization(net)
net = tflearn.activations.relu(net)
# Final layer weights are init to Uniform[-3e-3, 3e-3]
w_init = tflearn.initializations.uniform(minval=-0.003, maxval=0.003)
out = tflearn.fully_connected(
net, self.a_dim, activation='tanh', weights_init=w_init)
# Scale output to -action_bound to action_bound
scaled_out = tf.multiply(out, self.action_bound)
return inputs, out, scaled_out
What is the equivalent command for scaling the output layer according to my requirements?
python tensorflow keras deep-learning keras-rl
edited Nov 21 '18 at 1:19
Milo Lu
1,62711527
asked Nov 21 '18 at 0:39
CS101CS101
449
add a comment |
0
I am trying to implement DDPG from Keras RL and have the following actor network.
actor = Sequential()
actor.add(Flatten(input_shape=(1,) + env.observation_space.shape))
actor.add(Dense(16))
actor.add(Activation('relu'))
actor.add(Dense(16))
actor.add(Activation('relu'))
actor.add(Dense(16))
actor.add(Activation('relu'))
actor.add(Dense(nb_actions))
actor.add(Activation('linear'))
However, I would prefer to have the output scaled to a custom gym environment action space bounds for my problem. env.action_space.
https://pemami4911.github.io/blog/2016/08/21/ddpg-rl.html shows this using the tflearn api where they use
def create_actor_network(self):
inputs = tflearn.input_data(shape=[None, self.s_dim])
net = tflearn.fully_connected(inputs, 400)
net = tflearn.layers.normalization.batch_normalization(net)
net = tflearn.activations.relu(net)
net = tflearn.fully_connected(net, 300)
net = tflearn.layers.normalization.batch_normalization(net)
net = tflearn.activations.relu(net)
# Final layer weights are init to Uniform[-3e-3, 3e-3]
w_init = tflearn.initializations.uniform(minval=-0.003, maxval=0.003)
out = tflearn.fully_connected(
net, self.a_dim, activation='tanh', weights_init=w_init)
# Scale output to -action_bound to action_bound
scaled_out = tf.multiply(out, self.action_bound)
return inputs, out, scaled_out
What is the equivalent command for scaling the output layer according to my requirements?
python tensorflow keras deep-learning keras-rl
edited Nov 21 '18 at 1:19
Milo Lu
1,62711527
asked Nov 21 '18 at 0:39
CS101CS101
449
I do not want to write an answer but I realized I have to use a squashing function like tanh or sigmoid to map the network output to a range. Then define a function which will map the value from -1/+1 or 0/1 to the gym environment bounds. This is a more apt method for me since my custom gym environment bounds are not centered around 0
– CS101
Nov 21 '18 at 1:58
If my arguement is correct, can anyone tell me what kind of mapping should I define in my defined function. Shoulld I use proportional relationship between -1/+1 and my environment bounds or something more complex?
– CS101
Nov 21 '18 at 1:59
add a comment |
0
0
0
I am trying to implement DDPG from Keras RL and have the following actor network.
actor = Sequential()
actor.add(Flatten(input_shape=(1,) + env.observation_space.shape))
actor.add(Dense(16))
actor.add(Activation('relu'))
actor.add(Dense(16))
actor.add(Activation('relu'))
actor.add(Dense(16))
actor.add(Activation('relu'))
actor.add(Dense(nb_actions))
actor.add(Activation('linear'))
However, I would prefer to have the output scaled to a custom gym environment action space bounds for my problem. env.action_space.
https://pemami4911.github.io/blog/2016/08/21/ddpg-rl.html shows this using the tflearn api where they use
def create_actor_network(self):
inputs = tflearn.input_data(shape=[None, self.s_dim])
net = tflearn.fully_connected(inputs, 400)
net = tflearn.layers.normalization.batch_normalization(net)
net = tflearn.activations.relu(net)
net = tflearn.fully_connected(net, 300)
net = tflearn.layers.normalization.batch_normalization(net)
net = tflearn.activations.relu(net)
# Final layer weights are init to Uniform[-3e-3, 3e-3]
w_init = tflearn.initializations.uniform(minval=-0.003, maxval=0.003)
out = tflearn.fully_connected(
net, self.a_dim, activation='tanh', weights_init=w_init)
# Scale output to -action_bound to action_bound
scaled_out = tf.multiply(out, self.action_bound)
return inputs, out, scaled_out
What is the equivalent command for scaling the output layer according to my requirements?
python tensorflow keras deep-learning keras-rl
edited Nov 21 '18 at 1:19
Milo Lu
1,62711527
asked Nov 21 '18 at 0:39
CS101CS101
449
I am trying to implement DDPG from Keras RL and have the following actor network.
actor = Sequential()
actor.add(Flatten(input_shape=(1,) + env.observation_space.shape))
actor.add(Dense(16))
actor.add(Activation('relu'))
actor.add(Dense(16))
actor.add(Activation('relu'))
actor.add(Dense(16))
actor.add(Activation('relu'))
actor.add(Dense(nb_actions))
actor.add(Activation('linear'))
However, I would prefer to have the output scaled to a custom gym environment action space bounds for my problem. env.action_space.
https://pemami4911.github.io/blog/2016/08/21/ddpg-rl.html shows this using the tflearn api where they use
def create_actor_network(self):
inputs = tflearn.input_data(shape=[None, self.s_dim])
net = tflearn.fully_connected(inputs, 400)
net = tflearn.layers.normalization.batch_normalization(net)
net = tflearn.activations.relu(net)
net = tflearn.fully_connected(net, 300)
net = tflearn.layers.normalization.batch_normalization(net)
net = tflearn.activations.relu(net)
# Final layer weights are init to Uniform[-3e-3, 3e-3]
w_init = tflearn.initializations.uniform(minval=-0.003, maxval=0.003)
out = tflearn.fully_connected(
net, self.a_dim, activation='tanh', weights_init=w_init)
# Scale output to -action_bound to action_bound
scaled_out = tf.multiply(out, self.action_bound)
return inputs, out, scaled_out
What is the equivalent command for scaling the output layer according to my requirements?
python tensorflow keras deep-learning keras-rl
python tensorflow keras deep-learning keras-rl
edited Nov 21 '18 at 1:19
Milo Lu
1,62711527
asked Nov 21 '18 at 0:39
CS101CS101
449
edited Nov 21 '18 at 1:19
Milo Lu
1,62711527
asked Nov 21 '18 at 0:39
CS101CS101
449
edited Nov 21 '18 at 1:19
Milo Lu
1,62711527
edited Nov 21 '18 at 1:19
Milo Lu
1,62711527
edited Nov 21 '18 at 1:19
Milo Lu
1,62711527
1,62711527
asked Nov 21 '18 at 0:39
CS101CS101
449
asked Nov 21 '18 at 0:39
CS101CS101
449
asked Nov 21 '18 at 0:39
CS101CS101
449
449
I do not want to write an answer but I realized I have to use a squashing function like tanh or sigmoid to map the network output to a range. Then define a function which will map the value from -1/+1 or 0/1 to the gym environment bounds. This is a more apt method for me since my custom gym environment bounds are not centered around 0
– CS101
Nov 21 '18 at 1:58
If my arguement is correct, can anyone tell me what kind of mapping should I define in my defined function. Shoulld I use proportional relationship between -1/+1 and my environment bounds or something more complex?
– CS101
Nov 21 '18 at 1:59
add a comment |
I do not want to write an answer but I realized I have to use a squashing function like tanh or sigmoid to map the network output to a range. Then define a function which will map the value from -1/+1 or 0/1 to the gym environment bounds. This is a more apt method for me since my custom gym environment bounds are not centered around 0
– CS101
Nov 21 '18 at 1:58
If my arguement is correct, can anyone tell me what kind of mapping should I define in my defined function. Shoulld I use proportional relationship between -1/+1 and my environment bounds or something more complex?
– CS101
Nov 21 '18 at 1:59
I do not want to write an answer but I realized I have to use a squashing function like tanh or sigmoid to map the network output to a range. Then define a function which will map the value from -1/+1 or 0/1 to the gym environment bounds. This is a more apt method for me since my custom gym environment bounds are not centered around 0
– CS101
Nov 21 '18 at 1:58
I do not want to write an answer but I realized I have to use a squashing function like tanh or sigmoid to map the network output to a range. Then define a function which will map the value from -1/+1 or 0/1 to the gym environment bounds. This is a more apt method for me since my custom gym environment bounds are not centered around 0
– CS101
Nov 21 '18 at 1:58
If my arguement is correct, can anyone tell me what kind of mapping should I define in my defined function. Shoulld I use proportional relationship between -1/+1 and my environment bounds or something more complex?
– CS101
Nov 21 '18 at 1:59
If my arguement is correct, can anyone tell me what kind of mapping should I define in my defined function. Shoulld I use proportional relationship between -1/+1 and my environment bounds or something more complex?
– CS101
Nov 21 '18 at 1:59
add a comment |
0
active
oldest
votes
Your Answer
StackExchange.ifUsing("editor", function () {
StackExchange.using("externalEditor", function () {
StackExchange.using("snippets", function () {
StackExchange.snippets.init();
});
});
}, "code-snippets");
StackExchange.ready(function() {
var channelOptions = {
tags: "".split(" "),
id: "1"
};
initTagRenderer("".split(" "), "".split(" "), channelOptions);
StackExchange.using("externalEditor", function() {
// Have to fire editor after snippets, if snippets enabled
if (StackExchange.settings.snippets.snippetsEnabled) {
StackExchange.using("snippets", function() {
createEditor();
});
}
else {
createEditor();
}
});
function createEditor() {
StackExchange.prepareEditor({
heartbeatType: 'answer',
autoActivateHeartbeat: false,
convertImagesToLinks: true,
noModals: true,
showLowRepImageUploadWarning: true,
reputationToPostImages: 10,
bindNavPrevention: true,
postfix: "",
imageUploader: {
brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
allowUrls: true
},
onDemand: true,
discardSelector: ".discard-answer"
,immediatelyShowMarkdownHelp:true
});
}
});
draft saved
draft discarded
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fstackoverflow.com%2fquestions%2f53403722%2fscale-actor-network-output-to-the-action-space-bounds-in-keras-rl%23new-answer', 'question_page');
}
);
Post as a guest
Required, but never shown
0
active
oldest
votes
0
active
oldest
votes
active
oldest
votes
active
oldest
votes
draft saved
draft discarded
Thanks for contributing an answer to Stack Overflow!
- Please be sure to answer the question. Provide details and share your research!
But avoid …
- Asking for help, clarification, or responding to other answers.
- Making statements based on opinion; back them up with references or personal experience.
To learn more, see our tips on writing great answers.
draft saved
draft discarded
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fstackoverflow.com%2fquestions%2f53403722%2fscale-actor-network-output-to-the-action-space-bounds-in-keras-rl%23new-answer', 'question_page');
}
);
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
I do not want to write an answer but I realized I have to use a squashing function like tanh or sigmoid to map the network output to a range. Then define a function which will map the value from -1/+1 or 0/1 to the gym environment bounds. This is a more apt method for me since my custom gym environment bounds are not centered around 0
– CS101
Nov 21 '18 at 1:58
If my arguement is correct, can anyone tell me what kind of mapping should I define in my defined function. Shoulld I use proportional relationship between -1/+1 and my environment bounds or something more complex?
– CS101
Nov 21 '18 at 1:59