Questions tagged [deep-rl]

For questions related to deep reinforcement learning (DRL), that is, RL combined with deep learning. More precisely, deep neural networks are used to represent e.g. value functions or policies.

Filter by
Sorted by
Tagged with
15
votes
3answers
1k views

Why doesn't Q-learning converge when using function approximation?

The tabular Q-learning algorithm is guaranteed to find the optimal $Q$ function, $Q^*$, provided the following conditions (the Robbins-Monro conditions) regarding the learning rate are satisfied $\...
14
votes
1answer
11k views

How does LSTM in deep reinforcement learning differ from experience replay?

In the paper Deep Recurrent Q-Learning for Partially Observable MDPs, the author processed the Atari game frames with an LSTM layer at the end. My questions are: How does this method differ from the ...
14
votes
1answer
2k views

Why does DQN require two different networks?

I was going through this implementation of DQN and I see that on line 124 and 125 two different Q networks have been initialized. From my understanding, I think one network predicts the appropriate ...
8
votes
2answers
439 views

What is experience replay in laymen's terms?

I've been reading Google's DeepMind Atari paper and I'm trying to understand the concept of "experience replay". Experience replay comes up in a lot of other reinforcement learning papers (...
7
votes
3answers
1k views

How can you represent the state and action spaces for a card game in the case of a variable number of cards and actions?

I know how a machine can learn to play Atari games (Breakout): Playing Atari with Reinforcement Learning. With the same technique, it is even possible to play FPS games (Doom): Playing FPS Games with ...
7
votes
1answer
105 views

Why does reinforcement learning using a non-linear function approximator diverge when using strongly correlated data as input?

While reading the DQN paper, I found that randomly selecting and learning samples reduced divergence in RL using a non-linear function approximator (e.g a neural network). So why does Reinforcement ...
6
votes
2answers
385 views

Is reinforcement learning using shallow neural networks still deep reinforcement learning?

Often times I see the term deep reinforcement learning to refer to RL algorithms that use neural networks, regardless of whether or not the networks are deep. For example, PPO is often considered a ...
6
votes
1answer
1k views

Can TD($\lambda$) be used with deep reinforcement learning?

TD lambda is a way to interpolate between TD(0) - bootstrapping over a single step, and, TD(max), bootstrapping over the entire episode length, or, Monte Carlo. Reading the link above, I see that an ...
6
votes
2answers
741 views

Why is the log probability replaced with the importance sampling in the loss function?

In the Trust-Region Policy Optimisation (TRPO) algorithm (and subsequently in PPO also), I do not understand the motivation behind replacing the log probability term from standard policy gradients $$L^...
6
votes
2answers
576 views

Is it possible to implement reinforcement learning using a neural network?

I've implemented the reinforcement learning algorithm for an agent to play snappy bird (a shameless cheap ripoff of flappy bird) utilizing a q-table for storing the history for future lookups. It ...
5
votes
2answers
223 views

What are some online courses for deep reinforcement learning?

What are some (good) online courses for deep reinforcement learning? I would like the course to be both programming and theoretical. I really liked David Silver's course, but the course dates from ...
5
votes
2answers
3k 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 ...
5
votes
2answers
1k views

What is the difference between DQN and AlphaGo Zero?

I have already implemented a relatively simple DQN on Pacman. Now I would like to clearly understand the difference between a DQN and the techniques used by AlphaGo zero/AlphaZero and I couldn't find ...
5
votes
2answers
208 views

Is there an alternative to the use of target network?

In the context of Deep Q Network, a target network is usually utilized. The target network is a slow changing network with a changing rate as its hyperparameter. This includes both replacement update ...
5
votes
2answers
288 views

Is it possible to guide a reinforcement learning algorithm?

I have just started to study reinforcement learning and, as far as I understand, existing algorithms search for the optimal solution/policy, but do not allow the possibility for the programmer to ...
5
votes
2answers
332 views

Why don't people use projected Bellman error with deep neural networks?

Projected Bellman error has shown to be stable with linear function approximation. The technique is not at all new. I can only wonder why this technique is not adopted to use with non-linear function ...
5
votes
2answers
306 views

Can DQN perform better than Double DQN?

I'm training both DQN and double DQN in the same environment, but DQN performs significantly better than double DQN. As I've seen in the double DQN paper, double DQN should perform better than DQN. Am ...
5
votes
1answer
61 views

In imitation learning, do you simply inject optimal tuples of experience $(s, a, r, s')$ into your experience replay buffer?

Due to my RL algorithm having difficulties learning some control actions, I've decided to use imitation learning/apprenticeship learning to guide my RL to perform the optimal actions. I've read a few ...
4
votes
3answers
480 views

What is the target Q-value in DQNs?

I understand that in DQNs, the loss is measured by taking the MSE of outputted Q-values and target Q-values. Whats does the target Q-values represent? And how is it obtained/calculated by the DQN?
4
votes
2answers
422 views

Is there any good reference for double deep Q-learning?

I am new in reinforcement learning, but I already know deep Q-learning and Q-learning. Now, I want to learn about double deep Q-learning. Do you know any good references for double deep Q-learning? ...
4
votes
2answers
168 views

Why AlphaGo didn't use Deep Q-Learning?

In the previous research, in 2015, Deep Q-Learning shows its great performance on single player Atari Games. But why do AlphaGo's researchers use CNN + MCTS instead of Deep Q-Learning? is that because ...
4
votes
1answer
85 views

What does the notation $p_t = \text{max}_{i<t} p_i$ mean in algorithm 1 of the prioritized experience replay paper?

I am having a hard time converting line 6 of the prioritized experience replay algorithm from the original paper into plain English (see below): I understand that new transitions (not visited before) ...
4
votes
2answers
1k views

Do we have to use CNN for Deep Q Learning?

I read top articles on Google Search about Deep Q-Learning: https://medium.freecodecamp.org/an-introduction-to-deep-q-learning-lets-play-doom-54d02d8017d8 https://skymind.ai/wiki/deep-reinforcement-...
4
votes
1answer
189 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 & ...
4
votes
1answer
63 views

How can a single sample represent the expectation in gradient temporal difference learning?

I was reading the gradient temporal difference learning version 2(GTD2) from rich Sutton's book page-246. At some point, he expressed the whole expectation using a single sample from the environment. ...
4
votes
1answer
56 views

How does the repetition of features across states at different time steps affect learning?

Let's say you are training a neural network in an RL setting, where the state (i.e. features/input data) can be the same for multiple successive steps (~typically around 8 steps) of an episode. For ...
4
votes
1answer
64 views

How does policy evaluation work for continuous state space model-free approaches?

How does policy evaluation work for continuous state space model-free approaches? Theoretically, a model-based approach for the discrete state and action space can be computed via dynamic programming ...
4
votes
1answer
631 views

How is the gradient of the loss function in DQN derived?

In the original DQN paper, page 1, the loss function of the DQN is $$ L_{i}(\theta_{i}) = \mathbb{E}_{(s,a,r,s') \sim U(D)} [(r+\gamma \max_{a'} Q(s',a',\theta_{i}^{-}) - Q(s,a;\theta_{i}))^2] $$ ...
4
votes
0answers
341 views

What could be causing the drastic performance drop of the DQN model on the Pong environment?

I am running a basic DQN (Deep Q-Network) on the Pong environment. Not a CNN, just a 3 layer linear neural net with ReLUs. It seems to work for the most part, but at some point, my model suffers from ...
4
votes
0answers
105 views

How did the OpenAI 5 for Dota concatenate units?

I am no expert in the field of AI so I apologize if this is a simple/easy question. I was trying to implement a network similar to OpenAI's for another game and I noticed that I did not fully ...
3
votes
1answer
1k views

How should I model all available actions of a chess game in deep Q-learning?

I just read about deep Q-learning, which is using a neural network for the value function instead of a table. I saw the example here: https://yanpanlau.github.io/2016/07/10/FlappyBird-Keras.html and ...
3
votes
1answer
57 views

In the policy gradient equation, is $\pi(a_{t} | s_{t}, \theta)$ a distribution or a function?

I am learning about policy gradient methods from the Deep RL Bootcamp by Peter Abbeel and I am a bit stumbled by the math presented. In the lecture, he derives the gradient logarithm likelihood of a ...
3
votes
1answer
81 views

Why do authors track $\gamma_t$ in Prioritized Experience Replay Paper?

In the original prioritized experience replay paper, the authors track $\gamma_t$ in every state transition tuple (see line 6 in algorithm below): Why do the authors track this at every time step? ...
3
votes
2answers
1k views

My DQN is stuck and can't see where the problem is

I'm trying to replicate the DeepMind paper results, so I implemented my own DQN. I left it training for more than 4 million frames (more than 2000 episodes) on SpaceInvaders-v4 (OpenAI-Gym) and it ...
3
votes
1answer
58 views

Can AlphaZero considered as Multi-Agent Deep Reinforcement Learning?

Can AlphaZero considered as Multi-Agent Deep Reinforcement Learning? I could not find a clear answer on this. I would say yes it is Multi Agent Learning, as there are two Agents playing against each ...
3
votes
1answer
139 views

What is the difference between vanilla policy gradient and advantage actor-critic?

What is the difference between vanilla policy gradient (VPG) with a baseline as value function and advantage actor-critic (A2C)? By vanilla policy gradient I am specifically referring to spinning up's ...
3
votes
1answer
86 views

Are Q values estimated from a DQN different from a duelling DQN with the same number of layers and filters?

I am confused about the Q values of a duelling deep Q network (DQN). As far as I know, duelling DQNs have 2 outputs Advantage: how good it is to be in a particular state $s$ Value: the advantage of ...
3
votes
1answer
92 views

beautify an image with reinforcement learning

I am trying to formulate and solve the following problem of image mutation. Suppose I am trying to insert an object image into a "background" image of several objects, and I will need to look for a "...
3
votes
1answer
53 views

Maximizing or Minimizing in Trust Region Policy Optimization?

I happened to discover that the v1 (19 Feb 2015) and the v5 (20 Apr 2017) versions of TRPO papers have two different conclusions. The Equation (15) in v1 is $\min_\theta$ while the Equation (14) in v2 ...
3
votes
1answer
58 views

Can experience replay be used for training after completing every single epoch?

The DQN implements replay memory. Based on my research, I believe the replay memory starts to get used for training once there is enough experience in the memory buffer. This means the neural network ...
3
votes
1answer
63 views

Optimal RL function approximation for TicTacToe game

I modeled the TicTacToe game as a RL problem - with an environment and an agent. At first I made an "Exact" agent - using the SARSA algorithm, I saved every unique state, and chose the best (...
3
votes
1answer
62 views

How can I fix jerky movement in a continuous action space

I am training an agent to do object avoidance. The agent has control over its steering angle and its speed. The steering angle and speed are normalized in a $[−1,1]$ range, where the sign encodes ...
3
votes
1answer
46 views

Is there a logical method of deducing an optimal batch size when training a Deep Q-learning agent with experience replay?

I am training an RL agent using Deep-Q learning with experience replay. At each frame, I am currently sampling 32 random transitions from a queue which stores a maximum of 20000 and training as ...
3
votes
1answer
82 views

How to compute the target for double Q-learning update step?

I've already read the original paper about double DQN but I do not find a clear and practical explanation of how the target $y$ is computed, so here's how I interpreted the method (let's say I have 3 ...
3
votes
1answer
85 views

Can recovering a reward function using IRL lead to better policies compared to reward shaping?

I am working on a research project about the different reward functions being used in the RL domain. I have read up on Inverse Reinforcement Learning (IRL) and Reward Shaping (RS). I would like to ...
3
votes
1answer
74 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 ...
3
votes
1answer
342 views

What are the differences between the DQN variants?

There are several variants of the DQN model. For example, double DQN, duelling DQN, prioritized DQN, distributed prioritized DQN, episodic memory DQN, asynchronous n-step DQN and multiple DQN. What ...
3
votes
1answer
199 views

How does Hindsight Experience Replay learn from unsuccessful trajectories?

I am confused by how HER learns from unsuccessful trajectories. I understand that from failed trajectories it creates 'fake' goals that it can learn from. Ignoring HER for now, if in the case where ...
3
votes
1answer
759 views

Why use semi-gradient instead of full gradient in RL problems, when using function approximation?

Semi-gradient methods work well in reinforcement learning, but what is there a reason of not using the true gradient if it can be computed? I tried it on the cart pole problem with a deep Q-network ...
3
votes
0answers
15 views

DQN Agent with a 2D matrix as input in Keras

I have a Reinforcement Learning environment where the state is a 2D matrix with 0s and 1s (only one column with the value of 1 in each row). Example: ...