Questions tagged [ddpg]

For questions related to the reinforcement learning algorithm called Deep Deterministic Policy Gradient (DDPG).

Filter by
Sorted by
Tagged with
8 votes
2 answers
9k views

How large should the replay buffer be?

I'm learning DDPG algorithm by following the following link: Open AI Spinning Up document on DDPG, where it is written In order for the algorithm to have stable behavior, the replay buffer should ...
user avatar
7 votes
2 answers
696 views

Why are reinforcement learning methods sample inefficient?

Reinforcement learning methods are considered to be extremely sample inefficient. For example, in a recent DeepMind paper by Hessel et al., they showed that in order to reach human-level performance ...
user avatar
  • 253
7 votes
0 answers
1k views

Is there a difference in the architecture of deep reinforcement learning when multiple actions are performed instead of a single action?

I've built a deep deterministic policy gradient reinforcement learning agent to be able to handle any games/tasks that have only one action. However, the agent seems to fail horribly when there are ...
user avatar
  • 403
5 votes
1 answer
163 views

Why Q2 is a more or less independant estimate in Twin Delayed DDPG (TD3)?

Twin Delayed Deep Deterministic (TD3) policy gradient is inspired by both double Q-learning and double DQN. In double Q-learning, I understand that Q1 and Q2 are independent because they are trained ...
user avatar
4 votes
1 answer
2k views

How does the Ornstein-Uhlenbeck process work, and how it is used in DDPG?

In section 3 of the paper Continuous control with deep reinforcement learning, the authors write As detailed in the supplementary materials we used an Ornstein-Uhlenbeck process (Uhlenbeck & ...
user avatar
  • 41
4 votes
1 answer
622 views

What made your DDPG implementation on your environment work?

I am working on scheduling problem that has inherent randomness. The dimensions of action and state spaces are 1 and 5 respectively. I am using DDPG, but it seems extremely unstable, and so far it ...
user avatar
  • 242
4 votes
0 answers
333 views

What is the simplest policy gradient method to implement for a problem continuous action space?

I have a problem I would like to tackle with RL, but I am not sure if it is even doable. My agent has to figure out how to fill a very large vector (let's say from 600 to 4000 in the most complex ...
user avatar
  • 145
3 votes
1 answer
424 views

Appropriate algorithm for RL problem with sparse rewards, continuous actions and significant stochasticity

I'm working on a RL problem with the following properties: The rewards are extremely sparse i.e. all rewards are 0 except the terminal non-zero reward. Ideally I would not use any reward engineering ...
user avatar
  • 129
3 votes
1 answer
167 views

Purpose of using actor-critic algorithms under deterministic MDP dynamics?

One of the main disadvantages of the MC Policy Gradient algorithm (REINFORCE) as described say here is the fact that it has high variance (returns, which we sample, will significantly vary from ...
user avatar
  • 129
3 votes
1 answer
217 views

Training actor-critic algorithms in games with opponents

I am wondering how am I supposed to train a model using actor/critic algorithms in environments with opponents. I tried the followings (using A3C and DDPG): Play against random player. I had rather ...
user avatar
  • 266
3 votes
0 answers
182 views

How to deal with a moving target in the Lunar Lander environment with DDPG?

I have noticed that DDPG does rather well at solving environments with a static target. For example, the default of Lunar Lander, the flags do not change position. So the DDPG model learns how to get ...
user avatar
3 votes
1 answer
229 views

How to avoid rapid actuator movements in favor of smooth movements in a continuous space and action space problem?

I'm working on a continuous state / continuous action controller. It shall control a certain roll angle of an aircraft by issuing the correct aileron commands (in $[-1, 1]$). To this end, I use a ...
user avatar
  • 151
3 votes
0 answers
356 views

Should noise (such as OU) be decreased over time in actor / critic algorithms?

In most of RL algorithms I saw, there is a coefficient that reduces actions exploration over time, to help convergence. But in Actor-Critic, or other algorithms (A3C, DDPG, ...) used in continuous ...
user avatar
  • 266
3 votes
0 answers
121 views

Can I use deterministic policy gradient methods for stochastic policy learning?

Can I treat a stochastic policy (over a finite action space of size $n$) as a deterministic policy (in the set of probability distribution in $\mathbb{R}^n$)? It seems to me that nothing is broken ...
user avatar
2 votes
1 answer
191 views

Can I add expert data to the replay buffer used by the DDPG algorithm in order to make it converge faster?

I am working on a restricted reinforcement learning environment, i.e. the environment breaks very often (i.e.: the communication between the simulator and reinforcement learning agent breaks after ...
user avatar
2 votes
2 answers
667 views

Why is DDPG an off-policy RL algorithm?

In DDPG, if there are no $\epsilon$-greedy and no action noise, is DDPG an on-policy algorithm?
user avatar
2 votes
1 answer
69 views

How to sample the tuples during the initial time steps of the DDPG algorithm?

I am facing an issue in understanding the following line from the pseudocode of the DDPG algorithm Sample a random minibatch of $N$ transitions $(s_i, a_i, r_i, s_{i+1})$ from $R$ Here $N$ is a ...
user avatar
  • 3,201
2 votes
1 answer
228 views

Why does this Keras implementation of the DDPG algorithm update the critic's network using the gradient but the pseudocode doesn't?

I'm trying to understand the DDPG algorithm using Keras I found the site and started analyzing the code, I can't understand 2 things. The algorithm used to write the code presented on the page In the ...
user avatar
  • 21
2 votes
1 answer
59 views

Is there a good website where I can learn about Deep Deterministic Policy Gradient?

Is there a good website where I can learn about Deep Deterministic Policy Gradient?
user avatar
2 votes
1 answer
99 views

Why feed actions in later layer in Q network?

I read the DDPG paper, in which the authors state that the actions are fed only later to their Q network: Actions were not included until the 2nd hidden layer of Q. (Sec 7, Experiment Details) So ...
user avatar
  • 183
2 votes
1 answer
80 views

How does DDPG algorithm know about my action mapping function?

I am using DDPG to solve a RL problem. The action space is given by the Cartesian product $[0,20]^4\times[0,6]^4$. The actor is implemented as a deep neural network ...
user avatar
  • 247
2 votes
0 answers
591 views

Why is DDPG not learning and it does not converge?

I have used a different setting, but DDPG is not learning and it does not converge. I have used these codes 1,2, and 3 and I used different optimizers, activation functions, and learning rate but ...
user avatar
2 votes
0 answers
142 views

How can DDPG handle the discrete action space?

I am wondering how can DDPG or DPG handle the discrete action space. There are some papers saying that use Gumbel softmax with DDPG can make the discrete action problem be solved. However, will the ...
user avatar
  • 41
2 votes
0 answers
369 views

How does adding noise to the action in DDPG help in learning?

I can't understand how playing with the action generated by the actor network in DDPG by adding the noise term helps in exploration.
user avatar
2 votes
0 answers
75 views

How to learn using DDPG in python solely using a timeseries datasets

I have a lengthy timeseries datasets which contains several variables (from sensors etc) to be classified as actions or states. Providing they are successfully done, I want to learn a control policy ...
user avatar
  • 131
1 vote
1 answer
24 views

How to have zero value or a value between 200 and 400 in the output of a deep learning model?

I want to implement a DDPG method and obviously, the action space will be continuous. I have three outputs. The first output should be zero or a value between 200 and 400, and the other outputs have ...
user avatar
  • 11
1 vote
1 answer
149 views

Why is the policy loss the mean of $-Q(s, \mu(s))$ in the DDPG algorithm?

I am trying to implement the DDPG algorithm based on this paper. The part that confuses me is the actor network's update. I don't understand why the policy loss is simply the mean of $-Q(s, \mu(s))$, ...
user avatar
1 vote
2 answers
38 views

RL solutions for OpenAI Gym environments?

Is there any place where people share their agent's settings for solving OpenAI Gym Environments? For example, I'd like to know what are good parameters for a DDPG agent to learn the task in Reacher-...
user avatar
  • 41
1 vote
1 answer
88 views

Why is the behaviour policy denoted by $\beta$ and the exploration policy by $ \mu'$ in the DDPG paper?

I am learning about the deep deterministic policy gradient (DDPG) (Lillicrap et al, 2016) and got confused about the notation of the behavior policy. Lillicrap et al. denote the policy gradient by $$\...
user avatar
  • 35
1 vote
1 answer
79 views

What do the state features of KukaGymEnv represent? [closed]

I trying to use DDPG augmented with Hindsight Experience Replay (HER) on pybullet's KukaGymEnv. To formulate the feature vector for the goal state, I need to know what the features of the state of the ...
user avatar
1 vote
1 answer
879 views

Using DDPG for control in multi-dimensional continuous action space?

I am relatively new to reinforcement learning, and I am trying to implement a reinforcement learning algorithm that can do continuous control in a custom environment. The state of the environment is ...
user avatar
1 vote
1 answer
863 views

DDPG doesn't converge for MountainCarContinuous-v0 gym environment

I am trying to implement Deep Deterministic policy gradient algorithm by referring to the paper Continuous Control using Deep Reinforcement Learning on the MountainCarContinuous-v0 gym environment. I ...
user avatar
1 vote
1 answer
40 views

Are there examples of agents that use a more modest number of parameters on Pendulum (or similar environments)?

I'm looking at some baseline implementations of RL agents on the Pendulum environment. My guess was to use a relatively small neural net (~100 parameters). I'm comparing my solution with some ...
user avatar
  • 151
1 vote
1 answer
133 views

In Deep Deterministic Policy Gradient, are all weights of the policy network updated with the same or different value?

I'm trying to understand the DDPG algorithm shown at this page. I don't know what should the result of the gradient at step 14 be. Is it a scalar that I have to use to update all the weights (so all ...
user avatar
  • 291
1 vote
0 answers
18 views

Why can't I train like a dataset of samples instead of maintaining replay buffer?

On observing the DDPG algorithm, we notice that the updation of neural networks is happening during the episode. But, it seems there is no issue if we allow the completion of an episode and then treat ...
user avatar
  • 3,201
1 vote
0 answers
13 views

DDPG agent with convolutional layers for feature extraction [closed]

I'm trying to come up with a definition of the critic for a DDPG agent in PyTorch using a CNN as a feature extractor. It is pretty straight forward for the actor model. However, for the critic model I ...
user avatar
1 vote
0 answers
63 views

Could we add clipping in the output layer of the actor in DDPG?

I have a doubt about how clipping affects the training of the RL agents. In particular, I have come across a code for training DDPG agents, the pseudo-code is the following: ...
user avatar
  • 49
1 vote
0 answers
43 views

Setting initial values in DDPG to favor better actions

I'm working on a problem using DDPG. Is it possible to add some intelligence in the initialization phase, such that the convergence time is improved/shortened and local optima are avoided as much as ...
user avatar
  • 11
1 vote
0 answers
83 views

How to parallelize multi-agent DDPG (MADDPG)

I am experimenting with MADDPG algorithm implemented in this repo. Since there were only a few agents (2-3) in the implementation (also in the original paper) steps like parameter updates, action ...
user avatar
  • 197
1 vote
0 answers
85 views

Gradual decrease in performance of a DDPG agent

I'm trying to solve the OpenAI's CarRacing-v0 environment with the DDPG algorithm. I've observed that after a period of learning, the agent's performance starts to deteriorate slowly. For some ...
user avatar
1 vote
0 answers
162 views

How to deal with KerasRL DDPG algorithm getting stuck in a local optima?

I am using KerasRL DDPG to try to learn a policy on my own custom environment, but the agent is stuck in a local optima although I am adding the OrnsteinUhlenbeck randomization process. I used the ...
user avatar
  • 36
1 vote
0 answers
114 views

Why would DDPG with Hindsight Experience Replay not converge?

I am trying to train a DDPG agent augmented with Hindsight Experience Replay (HER) to solve the KukaGymEnv environment. The actor and critic are simple neural networks with two hidden layers (as in ...
user avatar
1 vote
0 answers
68 views

Benchmarking SAC on Pybullet

So far I have seen TD3 and DDPG benchmarks on Pybullet environments, but I am looking for SAC benchmarks on Pybullet too, anyone can help?
user avatar
  • 151
1 vote
0 answers
223 views

A question about the Wolpertinger algorithm (Deep RL in Large Discrete Action Spaces paper)

I am trying to reproduce the recommender task experiment from this paper. The paper suggests to embed discrete actions into continuous action space and then to use the proposed Wolpertinger agent. ...
user avatar
  • 11
1 vote
0 answers
28 views

What if the rewards induced by an environment are related to the policy too?

Assume we have a policy $\pi_{\theta}$ in a classic reinforcement learning setting, and a reward function $R^{\pi}(s,a)$ that changes as long as $\pi$ changes i.e. not only is it predefined by the ...
user avatar
  • 151
1 vote
0 answers
37 views

Why does the result when restoring a saved DDPG model differ significantly from the result when saving it?

I save the trained model after a certain number of episodes with the special save() function of the DDPG class (the network is saved when the reward reaches zero), but when I restore the model again ...
user avatar
1 vote
2 answers
166 views

Continuous control with DDPG: How to eliminate steady state error?

Currently I'm working on a continuous control problem using DDPG as my RL algorithm. All in all, things are working out quite well, but the algorithm does not show any tendencies to eliminate the ...
user avatar
  • 151
1 vote
0 answers
43 views

DDPG: how to implement continuous action space bounded in the interval [-2, 2]?

I am a newbie in reinforcement learning and trying to understand how to implement continuous actions bounded by $[-2, 2]$. My research shows that doing nothing is a possible solution (i.e. action of 4....
user avatar
  • 73
1 vote
0 answers
33 views

Should we multiply the target of actor by the importance sampling ratio when prioritized replay is applied to DDPG?

According to PER, we have to multiply the $Q$ error $\delta_i$ by the importance sampling ratio to correct the bias introduced by the imbalance sampling of PER, where importance sampling ratio is ...
user avatar
  • 431
0 votes
1 answer
318 views

Is DDPG just for deterministic environments?

I want to develop an AI for continuous space. I reached to DDPG algorithm that takes actions deterministically. If DDPG takes actions deterministically, should the environment also be deterministic? ...
user avatar
  • 107