Filters








337 Hits in 8.1 sec

Multiagent Deep Reinforcement Learning: Challenges and Directions Towards Human-Like Approaches [article]

Annie Wong, Thomas Bäck, Anna V. Kononova, Aske Plaat
2021 arXiv   pre-print
execution, opponent modelling, communication, efficient coordination, and reward shaping.  ...  We suggest that, for multiagent reinforcement learning to be successful, future research addresses these challenges with an interdisciplinary approach to open up new possibilities for more human-oriented  ...  Availability of data and material  ... 
arXiv:2106.15691v1 fatcat:7sy6cianq5dh5a7n6clvjdlrxy

State Elimination in Accelerated Multiagent Reinforcement Learning

Ary Setijadi Prihatmanto, Widyawardana Adiprawita, Safreni Candra Sari, Kuspriyanto Kuspriyanto
2016 International Journal on Electrical Engineering and Informatics  
This algorithm is generally applicable for other multiagent task challenges or general multiagent learning with large scale state space, and perfectly applicable with no adjustments for single agent learning  ...  without incorporating internal knowledge or human intervention such as reward shaping, transfer learning, parameter tuning, and even heuristics, into the learning system.  ...  This algorithm is not only applicable for primitive robot soccer task, but also for other robotic soccer task challenges with large scale state space.  ... 
doi:10.15676/ijeei.2016.8.3.12 fatcat:aydhzofht5hfhlypek4xkterpq

A Comprehensive Survey of Multiagent Reinforcement Learning

L. Busoniu, R. Babuska, B. De Schutter
2008 IEEE Transactions on Systems Man and Cybernetics Part C (Applications and Reviews)  
This paper provides a comprehensive survey of multiagent reinforcement learning (MARL). A central issue in the field is the formal statement of the multiagent learning goal.  ...  Additionally, the benefits and challenges of MARL are described along with some of the problem domains where the MARL techniques have been applied. Finally, an outlook for the field is provided.  ...  Together with the simplicity and generality of the setting, this makes RL attractive also for multiagent learning. However, several new challenges arise for RL in multiagent systems.  ... 
doi:10.1109/tsmcc.2007.913919 fatcat:tb2snaoxzbeothq2x27bg7tosa

A Survey and Critique of Multiagent Deep Reinforcement Learning [article]

Pablo Hernandez-Leal, Bilal Kartal, Matthew E. Taylor
2019 arXiv   pre-print
Recent works have explored learning beyond single-agent scenarios and have considered multiagent learning (MAL) scenarios.  ...  Deep reinforcement learning (RL) has achieved outstanding results in recent years. This has led to a dramatic increase in the number of applications and methods.  ...  expertise, to Baoxiang Wang for his suggestions on recent deep RL works, to Michael Kaisers, Daan Bloembergen, and Katja Hofmann for their comments about the practical challenges of MDRL, to the editor  ... 
arXiv:1810.05587v2 fatcat:h4ei5zx2xfa7xocktlefjrvef4

A Survey on Transfer Learning for Multiagent Reinforcement Learning Systems

Felipe Leno Da Silva, Anna Helena Reali Costa
2019 The Journal of Artificial Intelligence Research  
For this reason, reusing knowledge that can come from previous experience or other agents is indispensable to scale up multiagent RL algorithms.  ...  Multiagent Reinforcement Learning (RL) solves complex tasks that require coordination with other agents through autonomous exploration of the environment.  ...  This work was partially carried out while the first author was affiliated to the Learning Agents Research Group (LARG) at the University of Texas at Austin, TX, USA.  ... 
doi:10.1613/jair.1.11396 fatcat:mn4gw6oh5zgszl6l53fgesei5i

Muddling-Through and Deep Learning for Managing Large-Scale Uncertain Risks

Tony Cox
2019 Journal of Benefit-Cost Analysis  
We consider how recent insights from machine learning – especially, deep multiagent reinforcement learning – formalize aspects of muddling through and suggest principles for improving human organizational  ...  Deep learning principles adapted for human use can not only help participants in different levels of government or control hierarchies manage some large-scale distributed risks, but also show how rational-comprehensive  ...  Acknowledgments: I thank Vicki Bier and Warner North for valuable comments on an early draft that improved the framing, content and exposition in this paper.  ... 
doi:10.1017/bca.2019.17 fatcat:ch5f24amjnd7ndnptrwqw6zbo4

Evaluating the progress of Deep Reinforcement Learning in the real world: aligning domain-agnostic and domain-specific research [article]

Juan Jose Garau-Luis and Edward Crawley and Bruce Cameron
2021 arXiv   pre-print
Deep Reinforcement Learning (DRL) is considered a potential framework to improve many real-world autonomous systems; it has attracted the attention of multiple and diverse fields.  ...  Finally, we take up on ways to move forward accounting for both perspectives.  ...  learning Reward estima- tion or modifica- tion Try to overcome the challenge by directly modify- ing the reward structure and/or function Reward shaping in long trajectories, reward shaping for  ... 
arXiv:2107.03015v1 fatcat:ybp4mflut5hy7ikvychbwiuj4q

Can bounded and self-interested agents be teammates? Application to planning in ad hoc teams

Muthukumaran Chandrasekaran, Prashant Doshi, Yifeng Zeng, Yingke Chen
2016 Autonomous Agents and Multi-Agent Systems  
For the purposes of this study, individual, self-interested decision making in multiagent settings is modeled using interactive dynamic influence diagrams (I-DID).  ...  We address this limitation by including models at level 0 whose solutions involve reinforcement learning. We show how the learning is integrated into planning in the context of I-DIDs.  ...  Semi-model based learning in MCESP and other approximation techniques, will allow us to further scale-up augmented I-DIDs.  ... 
doi:10.1007/s10458-016-9354-4 fatcat:5vca5wsoenhsteassqszkla5oa

Distributed reinforcement learning for adaptive and robust network intrusion response

Kleanthis Malialis, Sam Devlin, Daniel Kudenko
2015 Connection science  
Such an attack is designed to exhaust a server's resources or congest a network's infrastructure, and therefore  ...  The increasing adoption of technologies and the exponential growth of networks has made the area of information technology an integral part of our lives, where network security plays a vital role.  ...  To better scale-up CTL is combined with a form of reward shaping.  ... 
doi:10.1080/09540091.2015.1031082 fatcat:vzwfb5cclzdqdhwgxzqozxe3xi

Reinforcement Learning-Empowered Mobile Edge Computing for 6G Edge Intelligence [article]

Peng Wei, Kun Guo, Ye Li, Jue Wang, Wei Feng, Shi Jin, Ning Ge, Ying-Chang Liang
2022 arXiv   pre-print
Furthermore, its evolved versions, such as deep RL (DRL), can achieve higher convergence speed efficiency and learning accuracy based on the parametric approximation for the large-scale state-action space  ...  Finally, the open challenges are discussed to provide helpful guidance for future research in RL training and learning MEC.  ...  Furthermore, to avoid the large-scale state space and action space, the multiagent Q-learning algorithm is employed for low-complexity task offloading.  ... 
arXiv:2201.11410v4 fatcat:24igkq4kbrb2pjzwf3mf3n7qtq

Reinforcement Learning Algorithms: Survey and Classification

N. R. Ravishankar, M. V. Vijayakumar
2017 Indian Journal of Science and Technology  
Under Reinforcement Learning it is very well known, there are 2 broad classifications as Model-based and Model-free RL 3 .  ...  Model-based RLs have the knowledge about the environment in which the agent acts, and about the agent, per se, as well.  ...  Bowling and Veloso's paper has examined a number of algorithms that solve multiagent RLs for Stochastic games, and it is noted that most of these algorithms, namely Shapely, PollatschekandAvi-Itzhak, Van  ... 
doi:10.17485/ijst/2017/v10i1/109385 fatcat:n7fsqz5mxfbetlopeplp2h2pke

Concurrent Meta Reinforcement Learning [article]

Emilio Parisotto and Soham Ghosh and Sai Bhargav Yalamanchi and Varsha Chinnaobireddy and Yuhuai Wu and Ruslan Salakhutdinov
2019 arXiv   pre-print
agent to reason over longer and longer time-scales.  ...  A negative side-effect of this sequential execution paradigm is that, as the environment becomes more and more challenging, and thus requiring more interaction episodes for the meta-learner, it needs the  ...  Acknowledgments: This work was supported in part by the Office of Naval Research grant #N000141812861, NSF Award IIS1763562, DARPA HR-00111890003, Apple, and Google focused award.  ... 
arXiv:1903.02710v1 fatcat:tmmhkxebo5fl5ilmjuzcy6lo3u

Deep Reinforcement Learning: A Brief Survey

Kai Arulkumaran, Marc Peter Deisenroth, Miles Brundage, Anil Anthony Bharath
2017 IEEE Signal Processing Magazine  
Currently, deep learning is enabling reinforcement learning to scale to problems that were previously intractable, such as learning to play video games directly from pixels.  ...  Deep reinforcement learning algorithms are also applied to robotics, allowing control policies for robots to be learned directly from camera inputs in the real world.  ...  Multiagent RL Usually, RL considers a single learning agent in a stationary environment.  ... 
doi:10.1109/msp.2017.2743240 fatcat:lslmokzwirgilalorttl443524

Quantifying the impact of non-stationarity in reinforcement learning-based traffic signal control

Lucas N. Alegre, Ana L.C. Bazzan, Bruno C. da Silva
2021 PeerJ Computer Science  
Causes for and effects of this are manifold.  ...  In reinforcement learning (RL), dealing with non-stationarity is a challenging issue. However, some domains such as traffic optimization are inherently non-stationary.  ...  We evaluate the performance of deploying RL in a non-stationary multiagent scenario, where each traffic signal uses Q-learning-a model-free RL algorithm-to learn efficient control policies.  ... 
doi:10.7717/peerj-cs.575 pmid:34141896 pmcid:PMC8176548 fatcat:ptng2mzyfzcarkchupz4ii7che

Reinforcement Learning Assisted Beamforming for Inter-cell Interference Mitigation in 5G Massive MIMO Networks [article]

Aidong Yang, Xinlang Yue, Ye Ouyang
2021 arXiv   pre-print
In this paper, we propose a reinforcement learning (RL) assisted full dynamic beamforming for ICI mitigation in 5G downlink.  ...  The proposed algorithm is a joint of beamforming and full dynamic Q-learning technology to minimize the ICI, and results in a low-complexity method without channel estimation.  ...  Let initial function of be ( , ) = 0 ( , ) with shaping rewards Φ( ′ ) − Φ( ), and initial function of ′ be ′ ( , ) = 0 ( , ) + Φ( ) with no shaping rewards.  ... 
arXiv:2103.11782v2 fatcat:4g4qei4v7besbgba5whz6gcs6e
« Previous Showing results 1 — 15 out of 337 results