Questions tagged [reinforcement-learning]

For questions related to learning controlled by external positive reinforcement or negative feedback signal or both, where learning and use of what has been thus far learned occur concurrently.

223 questions with no upvoted or accepted answers
Filter by
Sorted by
Tagged with
5
votes
1answer
106 views

How do I compute the variance of the return of an evaluation policy using two behaviour policies?

Suppose there is an evaluation policy called $\pi_{e}$ and there are two behavior policies $\pi_{b1}$ and $\pi_{b2}$. I know that it is possible to estimate the return of policy $\pi_{e}$ through ...
5
votes
1answer
211 views

Building AI from chess - data shape from simulation

Problem My problem is the following: Given 1000 wins, losses, and ties from a chess simulation I am using, what shape should each game be (I.e., sequence of moves leading to win/loss/tie) in order to ...
5
votes
2answers
433 views

Reinforcement Learning with asynchronous feedback

I want suggestions on literature on Reinforcement Learning algorithms that perform well with asynchronous feedback from the environment. What I mean by asynchronous ...
4
votes
0answers
54 views

Unable to train Coach for Banana-v0 Gym environment

I have just started playing with Reinforcement learning and starting from the basics I'm trying to figure out how to solve Banana Gym with coach. Essentially ...
4
votes
0answers
44 views

Normalizing Normal Distributions in Thompson Sampling for online Reinforcement Learning

In my implementation of Thompson sampling (TS) for online Reinforcement Learning, my distribution for selecting $a$ is $\mathcal{N}(Q(s, a), \frac{1}{C(s,a)+1})$ where $C(s,a)$ is the number of times $...
4
votes
0answers
32 views

Counterexamples to the reward hypothesis

On Sutton and Barto's RL book, the reward hypothesis is stated as that all of what we mean by goals and purposes can be well thought of as the maximization of the expected value of the cumulative ...
4
votes
2answers
102 views

Are there any online competitions for Reinforcement Learning?

Kaggle is limited to only supervised learning problems. There used to be www.rl-competition.org but they've stopped. Is there anything else I can do other than locally trying out different algorithms ...
4
votes
0answers
124 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
53 views

Inconsistent definitions of the retrace

In Section 4.3 of paper Learning by Playing - Solving Sparse Reward Tasks from Scratch, the authors define Retrace as $$ Q^{ret}=\sum_{j=i}^\infty\left(\gamma^{j-i}\prod_{k=i}^jc_k\right)[r(s_j,a_j)+\...
4
votes
0answers
34 views

What research has been done on learning non-Markovian reward functions?

Recently, some work has been done planning and learning in Non-Markovian Decision Processes, that is, decision-making with temporally extended rewards. In these settings, a particular reward is ...
4
votes
0answers
79 views

Is there any theoretical capabilities of apply deep successor representations with A3C algorithm?

Deep Successor Representations(DSR) has given better performance in tasks like navigation when it compares to normal model-free RL tasks. Basically, DSR is a hybrid of model-free RL and Model-Based RL....
4
votes
1answer
738 views

Traveling salesman problem variant: which algorithm to choose?

I have an industrial problem which I'm trying to cast as a Traveling Salesman problem (TSP) in 3D euclidian space. There are physical limitations which implies that some subpaths may or may not be ...
3
votes
0answers
23 views

Ideas on a network that can translate image differences into motor commands?

I'd like to design a network that gets two images (an image under construction, and an ideal image), and has to come up with an action vector for a simple motor command which would augment the image ...
3
votes
1answer
46 views

Convergence of Semi gradient TD(0) with non-linear function approximation

I am looking for a result that shows convergence of semi-gradient TD(0) algorithm with non-linear function approximation for on-policy prediction. Specifically, the update equation is given by (...
3
votes
1answer
30 views

What could be the cause of the drop in the reward in A3C?

The mean episodic reward is generally increasing, but it has spontaneous drops, and I'm not sure of their cause. The problem has a sparse reward, batch size=2000, <...
3
votes
0answers
25 views

What is the current state-of-the-art for inverse reinforcement learning on robotic control tasks?

The question is broad, since I am looking for any kind of insight into the subject. I am currently looking into the possibility for useing inverse reinforcement learning for learning robotic ...
3
votes
0answers
32 views

Is there any open source implementation of the SBEED learning algorithm?

Are there are any openly available implementations of the SBEED: Convergent Reinforcement Learning with Nonlinear Function Approximation paper?
3
votes
1answer
73 views

Sampling in TRPO or PPO

In the TRPO paper, the objective to maximize is (equation 14) $$ \mathbb{E}_{s\sim\rho_{\theta_\text{old}},a\sim q}\left[\frac{\pi_\theta(a|s)}{q(a|s)} Q_{\theta_\text{old}}(s,a) \right] $$ which ...
3
votes
0answers
35 views

What is the difference between random and sequential sampling from the reply memory?

I was working on an RL problem and I am confused at one specific point. We use replay memory so that the network learns about previous actions and how these actions lead to a success or a failure. ...
3
votes
0answers
81 views

How can I use Q-learning for inventory decision making?

I am trying to model operational decisions in inventory control. The control policy is base stock with a fixed stock level of $S$. That is replenishment orders are placed for every demand arrival to ...
3
votes
2answers
69 views

What is the difference between return and expected return?

At a time step $t$, for a state $S_{t}$, the return is defined as the discounted cumulative reward from that time step $t$. If an agent is following a policy (which in itself is a probability ...
3
votes
0answers
63 views

How to represent action space in reinforcement learning?

I started to learn reinforcement learning a few days ago. And I want to use that to solve resource allocation problem something like given a constant number, find the best way to divide it into ...
3
votes
0answers
25 views

Can $\Phi$ measure of Integrated Information Theory serve as reward for reinforcement learning system?

Can $\Phi$ measure (computed rigorously or approximately) of Integrated Information Theory serve as reward for self-evolving/learning reinforcement learning system and hence we let this system to ...
3
votes
0answers
23 views

Humanoid agent reward shaping

I'm trying to teach a humanoid agent how to stand up after falling. The episode starts with the agent lying on the floor with its back touching the ground, and its goal is to stand up in the shortest ...
3
votes
0answers
47 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 ...
3
votes
0answers
29 views

Code examples of controlling multiple units with RL

Anyone knows a resources (papers, articles and especially repositories) regarding controlling multiple units with RL. The controlled units should not be fixed, for example in Real Time Strategy the ...
3
votes
0answers
41 views

Deep Q-Learning agent poor performing actions. Need help optimizing

I'm trying to make deep q-learning agent from https://keon.io/deep-q-learning My environment looks like this: https://imgur.com/a/OnbiCtV As you can see my agent is a circle and there is one gray ...
3
votes
0answers
48 views

Feature Selection using Monte Carlo Tree Search

I'm trying to tackle the problem of feature selection as an RL problem, inspired by the paper Feature Selection as a One-Player Game. I know Monte-Carlo tree search (MCTS) is hardly RL. So, I used ...
3
votes
0answers
110 views

What can be considered a deep recurrent neural network?

In the paper Deep Recurrent Q-Learning for Partially Observable MDPs, the DRQN is described as DQN with the first post-convolutional fully-connected layer replaced by a recurrent LSTM. I have DQN ...
3
votes
0answers
69 views

What is a generalized MDP?

What is a generalized MDP? How is it different than a "regular" MDP? How does it generalise the notion of an MDP? Why do we need a generalised MDP? Do generalised MDPs have some practical usefulness ...
3
votes
0answers
37 views

Negative counterfactual regret

I am reading the paper Regret Minimization in Games with Incomplete Information on CFR algorithm. On page 4, the paper defines $R^{T,+}_{i,\text{imm}}=\max\{R^{T}_{i,\text{imm}}, 0\}$ after equation (...
3
votes
1answer
73 views

Can A3C update the policy / critic on a local machine without needing to copy?

To make A2C into A3C you make it asynchronous. From what I understand the 'correct' way to do that is to thread off workers with a copy of the policy and critic, and then return the state/action/...
3
votes
0answers
43 views

Choosing more than one action in a parameterized policy

I would like to implement a variant of policy iteration that can choose one or more actions in each state. An example would be to heal and move in the game of Doom. Parameterizing the power set of ...
3
votes
3answers
510 views

How to implement a Continuous Control of a quadruped robot with Deep Reinforcement Learning in Pybullet and OpenAI Gym?

Description I have designed this robot in URDF format and its environment in pybullet. Each leg has a minimum and maximum value of movement. What reinforcement algorithm will be best to create a ...
3
votes
0answers
92 views

Why are all the actions converging to the same index?

I am using PPO with an LSTM agent. My agent is performing 10 actions for each episode, one action is corresponding to one LSTM timestep and the action space is discrete. I have only one reward per ...
3
votes
0answers
172 views

Understanding multi iteration update of model in Policy Gradient PPO algorithm

I have a general question about the updating of the network/model in the PPO algorithm. If I understand it correctly, there are multiple iterations of weight updates done on the model with data that ...
3
votes
0answers
372 views

RL to generate sentences

I want to develop a system to generate grammatically correct sentences. The input would be some words. The output would be a grammatically correct human-like sentence. Eg: Input: capital, Paris, ...
2
votes
0answers
43 views

What form of output would be needed to train a model on a connect 4 AI?

I've had a big interest in machine learning for a while, and I've followed along a few tutorials, but have never made my own project. After losing many games of connect 4 with my friends, I decided to ...
2
votes
0answers
24 views

How would one develop an action space for a game that is proprietary?

I'm currently trying to develop an RL that will teach itself to play the popular fighting game "Tekken 7". I initially had the idea of teaching it to play generally- against actual opponents with ...
2
votes
1answer
41 views

Understanding proof of lemma 1 (policy improvement bound) of the “Trust Region Policy Optimization” paper

In the Trust Region Policy Optimization paper, in Lemma 1 of Appendix A, I did not quite understand the transition from (21) from (20). In going from (20) to (21), $A^\pi(s_t, a_t)$ is substituted ...
2
votes
0answers
23 views

What could be the cause of the drop of the total reward when using DQN to solve the cart-pole environment?

I'm trying to use DQN to solve the cart-pole environment. I have 2 networks (target and behavior). Both of them have 3 hidden layers with 24 neurons, using the ReLU activation. The loss is MSE and the ...
2
votes
0answers
23 views

How am I getting same results 30 times faster than in original HER paper?

I am reproducing the results from Hindsight Experience Replay by Andrychowicz et. al. In the original paper they present the results below, where the agent is trained for 200 epochs. 200 epochs * 800 ...
2
votes
0answers
9 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 ...
2
votes
1answer
42 views

REINFORCE algorithm for portfolio optimization - problem while training

I'm trying to implement the Reinforce algorithm (Monte Carlo policy gradient) in order to optimize a portfolio of 94 stocks on a daily basis (I have suitable historical data to achieve this). The idea ...
2
votes
0answers
28 views

Best approach for online Machine Translation with few hundred of samples?

I want to implement a model that improves itself with the passage of time. My main task is to build a machine translator (from English to Urdu).. The problem I am facing is that I have very little ...
2
votes
0answers
28 views

Is there a way to do reinforcement learning in POMDP?

Are there any algorithms to use reinforcement learning to learn optimal policies in partially observable Markov decision process (POMDP) i.e. when the state is not perfectly observed. More ...
2
votes
0answers
31 views

how to use Softmax action selection algorithm in atari-like game

I'm currently writing a program using keras (python 3) to play a game similar to Atari games, only in this one there are objects moving in the screen in different angles and directions (in most of ...
2
votes
0answers
37 views

Feasibility of using machine learning to obtain self-consistent solutions

I am a physicist and I don't have much background on machine learning or deep learning except taking a couple of courses on statistics. In physics, we often simulate a model by means of two-way ...
2
votes
0answers
27 views

Should importance sample weighting be compensated for by dynamically increasing learning rate?

I'm using Prioritized Experience Replay (PER) with a DDQN. To compensate for overfitting relatively high-value samples due to the non-uniform selection, I'm training with sample weights provided along ...
2
votes
0answers
47 views

Torch CNN not training

I am completely new to CNN's, and I do not quite know how to design or use them efficiently. That being said, I am attempting to build a CNN that learns to play Pac-man with reinforcement learning. I ...