Filters








1,377 Hits in 9.2 sec

Heuristically Accelerated Q–Learning: A New Approach to Speed Up Reinforcement Learning [chapter]

Reinaldo A. C. Bianchi, Carlos H. C. Ribeiro, Anna H. R. Costa
2004 Lecture Notes in Computer Science  
This work presents a new algorithm, called Heuristically Accelerated Q-Learning (HAQL), that allows the use of heuristics to speed up the well-known Reinforcement Learning algorithm Q-learning.  ...  Finally, experimental results shows that even a very simple heuristic results in a significant enhancement of performance of the reinforcement learning algorithm.  ...  Conclusion and Future Works This work presented a new algorithm, called Heuristically Accelerated Q-Learning (HAQL), that allows the use of heuristics to speed up the well-known Reinforcement Learning  ... 
doi:10.1007/978-3-540-28645-5_25 fatcat:mb6h7ni6tjgpjdvzvnbtzpubai

Distributed Heuristically Accelerated Q-Learning for Robust Cognitive Spectrum Management in LTE Cellular Systems

Nils Morozs, Tim Clarke, David Grace
2016 IEEE Transactions on Mobile Computing  
In addition, DIAQ dramatically improves initial performance, speeds up convergence and improves steady state performance of a state-of-the-art distributed Q-learning algorithm, confirming the theoretical  ...  It combines distributed reinforcement learning (RL) and standardized inter-cell interference coordination (ICIC) signalling in the LTE downlink, using the framework of heuristically accelerated RL (HARL  ...  An emerging technique to mitigate this poor initial performance problem is the heuristically accelerated reinforcement learning (HARL) approach, where additional heuristic information is used to guide  ... 
doi:10.1109/tmc.2015.2442529 fatcat:ilhxr2jumzdmnfkjh2gymauxdq

Accelerating autonomous learning by using heuristic selection of actions

Reinaldo A. C. Bianchi, Carlos H. C. Ribeiro, Anna H. R. Costa
2007 Journal of Heuristics  
The paper is organized as follows: Section 2 briefly reviews the reinforcement learning approach, describes the Q-Learning algorithm, and presents some approaches to speeding up RL.  ...  Approaches to speed up Reinforcement Learning The SARSA Algorithm (Sutton, 1996) is a modification of Q-learning that admits the next action to be chosen randomly according to a predefined probability,  ... 
doi:10.1007/s10732-007-9031-5 fatcat:qqj5xz63c5akfoxx4f6e7pltga

Dynamic heuristic acceleration of linearly approximated SARSA( $$\lambda $$ λ ): using ant colony optimization to learn heuristics dynamically

Stefano Bromuri
2019 Journal of Heuristics  
Heuristically accelerated reinforcement learning (HARL) is a new family of algorithms that combines the advantages of reinforcement learning (RL) with the advantages of heuristic algorithms.  ...  The proposed dynamic algorithms are evaluated in comparison to linearly approximated SARSA(λ), and heuristically accelerated SARSA(λ) using a static heuristic in three benchmark scenarios: the mountain  ...  the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.  ... 
doi:10.1007/s10732-019-09408-x fatcat:63vhqqr6erci3beb6cxtr7qtme

Heuristically-Accelerated Reinforcement Learning: A Comparative Analysis of Performance [chapter]

Murilo Fernandes Martins, Reinaldo A. C. Bianchi
2014 Lecture Notes in Computer Science  
This paper presents a comparative analysis of three Reinforcement Learning algorithms (Q-learning, Q(λ)-learning and QSlearning) and their heuristically-accelerated variants (HAQL, HAQ(λ) and HAQS) where  ...  heuristics bias action selection, thus speeding up the learning.  ...  Heuristically-Accelerated Reinforcement Learning The use of heuristics to accelerate RL algorithms has firstly been proposed in [6] , where the Q-learning algorithm was extended to take advantage of a  ... 
doi:10.1007/978-3-662-43645-5_2 fatcat:i6ayghqymncppdb4ujyil3k5zi

Heuristically Accelerated Reinforcement Learning for Dynamic Secondary Spectrum Sharing

Nils Morozs, Tim Clarke, David Grace
2015 IEEE Access  
In this paper we propose a heuristically accelerated reinforcement learning (HARL) based framework, designed for dynamic secondary spectrum sharing in LTE cellular systems.  ...  This results in a significant decrease in primary system quality of service degradation due to the interference from the secondary cognitive systems, compared to a state-of-the-art reinforcement learning  ...  One of the more recent promising solutions to this issue, proposed in the artificial intelligence domain, is the heuristically accelerated reinforcement learning (HARL) approach.  ... 
doi:10.1109/access.2015.2507158 fatcat:iwbs4bz535bejitrsdg37ptoqm

Efficient Use of heuristics for accelerating XCS-based Policy Learning in Markov Games [article]

Hao Chen, Chang Wang, Jian Huang, Jianxing Gong
2020 arXiv   pre-print
This paper proposes efficient use of rough heuristics to speed up policy learning when playing against concurrent learners.  ...  Furthermore, we introduce an accuracy-based eligibility trace mechanism to speed up rule evolution, i.e., classifiers that can match the historical traces are reinforced according to their accuracy.  ...  To alleviating this problem and speed up the learning process, a variety of accelerating techniques and corresponding Q-table based RL (QbRL) algorithms have been proposed.  ... 
arXiv:2005.12553v1 fatcat:g3jowlii6badtizyikxxfxyk5y

Constrained Deep Q-Learning Gradually Approaching Ordinary Q-Learning

Shota Ohnishi, Eiji Uchibe, Yotaro Yamaguchi, Kosuke Nakanishi, Yuji Yasui, Shin Ishii
2019 Frontiers in Neurorobotics  
A deep Q network (DQN) (Mnih et al., 2013) is an extension of Q learning, which is a typical deep reinforcement learning method.  ...  In the proposed method, as learning progresses, the number of times that the constraints are activated decreases. Consequently, the update method gradually approaches conventional Q learning.  ...  rule approaches that of ordinary Q learning.  ... 
doi:10.3389/fnbot.2019.00103 pmid:31920613 pmcid:PMC6914867 fatcat:b7ujdkvl3zc7jbgdrqqglnyplu

Hyperparameter Optimization for Tracking with Continuous Deep Q-Learning

Xingping Dong, Jianbing Shen, Wenguan Wang, Yu Liu, Ling Shao, Fatih Porikli
2018 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition  
To overcome this challenge, we introduce an efficient heuristic to accelerate the convergence behavior. We evaluate our method on several tracking benchmarks and demonstrate its superior performance 1  ...  Here, we propose a novel hyperparameter optimization method that can find optimal hyperparameters for a given sequence using an action-prediction network leveraged on Continuous Deep Q-Learning.  ...  [31] propose a simple model to enable deep Q-learning into the continuous action space and accelerate learning with new imagination rollouts.  ... 
doi:10.1109/cvpr.2018.00061 dblp:conf/cvpr/DongSWL0P18 fatcat:vfngzqqk7bedhnvf4nbpce6tji

Natural Gradient Deep Q-learning [article]

Ethan Knight, Osher Lerner
2018 arXiv   pre-print
We present a novel algorithm to train a deep Q-learning agent using natural-gradient techniques.  ...  Together these results suggest that natural-gradient techniques can improve value-function optimization in deep reinforcement learning.  ...  Q-learning Q-learning [Watkins, 1989, Rummery and Niranjan, 1994 ] is a model-free reinforcement learning algorithm which works by gradually learning Q(s, a), the expectation of the cumulative reward.  ... 
arXiv:1803.07482v2 fatcat:mr3kfotoozdgtflsqzhuzxgwsm

Multi-Agent Advisor Q-Learning [article]

Sriram Ganapathi Subramanian, Matthew E. Taylor, Kate Larson, Mark Crowley
2022 arXiv   pre-print
An interesting question that arises is how to best use such approaches as advisors to help improve reinforcement learning in multi-agent domains.  ...  However, many real-world environments already, in practice, deploy sub-optimal or heuristic approaches for generating policies.  ...  (Amii); a Canada CIFAR AI Chair, Amii; Compute Canada; Mitacs; and NSERC.  ... 
arXiv:2111.00345v5 fatcat:o3ybmwuuubebnn4ntqfxrhwzjy

Continuous Deep Q-Learning with Model-based Acceleration [article]

Shixiang Gu and Timothy Lillicrap and Ilya Sutskever and Sergey Levine
2016 arXiv   pre-print
To further improve the efficiency of our approach, we explore the use of learned models for accelerating model-free reinforcement learning.  ...  NAF representation allows us to apply Q-learning with experience replay to continuous tasks, and substantially improves performance on a set of simulated robotic control tasks.  ...  Beyond deriving an improved model-free deep reinforcement learning algorithm, we also seek to incorporate elements of model-based RL to accelerate learning, without giving up the strengths of model-free  ... 
arXiv:1603.00748v1 fatcat:jdrmstwwm5au7lrper5r5dicum

Multi-Agent Advisor Q-Learning

Sriram Ganapathi Subramanian, Matthew E. Taylor, Kate Larson, Mark Crowley
2022 The Journal of Artificial Intelligence Research  
An interesting question that arises is how to best use such approaches as advisors to help improve reinforcement learning in multi-agent domains.  ...  However, many real-world environments already, in practice, deploy sub-optimal or heuristic approaches for generating policies.  ...  (Amii); a Canada CIFAR AI Chair, Amii; Compute Canada; Mitacs; and NSERC.  ... 
doi:10.1613/jair.1.13445 fatcat:tgvw3lf62bdp5jlnxrakeszw5i

Autonomous Driving in Roundabout Maneuvers Using Reinforcement Learning with Q-Learning

Laura García Cuenca, Enrique Puertas, Javier Fernandez Andrés, Nourdine Aliane
2019 Electronics  
This paper proposes an approach based on the use of the Q-learning algorithm to train an autonomous vehicle agent to learn how to appropriately navigate roundabouts.  ...  The results illustrate that the Q-learning-algorithm-based vehicle agent is able to learn smooth and efficient driving to perform maneuvers within roundabouts.  ...  Currently, several machine learning algorithms use the reinforcement learning paradigm as the basis for implementation, such as adaptive heuristic critic (AHC) [13] , Q-learning [14] , Markov decision  ... 
doi:10.3390/electronics8121536 fatcat:a7dod2evqbddnlvazefknzqrry

A Behavior-based Approach for Multi-agent Q-learning for Autonomous Exploration [article]

Dip Narayan Ray, Somajyoti Majumder, Sumit Mukhopadhyay
2011 arXiv   pre-print
This paper describes such a multiagent approach for implementing a type of reinforcement learning using a priority based behaviour-based architecture.  ...  Currently reinforcement learning is getting more acceptances for implementing learning in robots from the system-environment interactions.  ...  This algorithm is based upon an emerging technique, Heuristic Accelerated Reinforcement Learning, in which RL methods are accelerated by making use of heuristic information.  ... 
arXiv:1110.1796v1 fatcat:fdla7znvkvcdnjams4c26orfoa
« Previous Showing results 1 — 15 out of 1,377 results