Filters








57,592 Hits in 2.5 sec

Learning to Resolve Alliance Dilemmas in Many-Player Zero-Sum Games [article]

Edward Hughes, Thomas W. Anthony, Tom Eccles, Joel Z. Leibo, David Balduzzi, Yoram Bachrach
2020 arXiv   pre-print
Using symmetric zero-sum matrix games, we demonstrate formally that alliance formation may be seen as a social dilemma, and empirically that na\"ive multi-agent reinforcement learning therefore fails to  ...  We introduce a toy model of economic competition, and show how reinforcement learning may be augmented with a peer-to-peer contract mechanism to discover and enforce alliances.  ...  Despite the fact that Matching yields a strict social dilemma while Odd One Out gives a non-strict social dilemma, the learning dynamics in both cases fail to find the optimal alliances.  ... 
arXiv:2003.00799v1 fatcat:evsc7hmlvndbrppcfwrnwzcvaq

Numerical analysis of a reinforcement learning model with the dynamic aspiration level in the iterated Prisoner's dilemma

Naoki Masuda, Mitsuhiro Nakamura
2011 Journal of Theoretical Biology  
It may serve to explore the relationships between learning and evolution in social dilemma situations.  ...  In the context of the iterated Prisoner's Dilemma, we numerically examine the performance of a reinforcement learning model.  ...  N.M. acknowledges the support from the Grants-in-Aid for Scientific Research (No. 20760258). M.N. acknowledges the support and the Grants-in-Aid for Scientific Research from JSPS, Japan.  ... 
doi:10.1016/j.jtbi.2011.03.005 pmid:21397610 fatcat:vykschk6svg3ff6ynwnxljnaly

Learning dynamics in social dilemmas

M. W. Macy, A. Flache
2002 Proceedings of the National Academy of Sciences of the United States of America  
Using a BM stochastic learning model, we identify a fundamental solution concept for the long-term dynamics of backward-looking behavior in all social dilemmas-stochastic collusion-based on random walk  ...  In social dilemmas, if aspiration levels are below maximin, then mutual or unilateral defection may also be mutually reinforcing, even though these outcomes are socially deficient.  ... 
doi:10.1073/pnas.092080099 pmid:12011402 pmcid:PMC128590 fatcat:rgove6gwkze65ned7ed2654yse

Improved cooperation by balancing exploration and exploitation in intertemporal social dilemma tasks [article]

Zhenbo Cheng, Xingguang Liu, Leilei Zhang, Hangcheng Meng, Qin Li, Xiao Gang
2021 arXiv   pre-print
We demonstrate that agents that use the simple strategy improve a relatively collective return in a decision task called the intertemporal social dilemma, where the conflict between the individual and  ...  We also explore the effects of the diversity of learning rates on the population of reinforcement learning agents and show that agents trained in heterogeneous populations develop particularly coordinated  ...  In Figure 2 , we compare the collective return of groups performing inter-period social dilemma tasks for a fixed learning rate and dynamic learning rate.  ... 
arXiv:2111.09152v1 fatcat:h6s4it4yyraehmtg2ujtl5jofu

A model for the evolution of reinforcement learning in fluctuating games

Slimane Dridi, Laurent Lehmann
2015 Animal Behaviour  
We analyse through stochastic approximation theory and simulations the learning dynamics on the behavioural timescale, and derive conditions where trial-and-error learning outcompetes hypothetical reinforcement  ...  Here, we develop an evolutionary model in which individuals are genetically determined to use either trial-and-error learning or learning based on hypothetical reinforcements, and ask what is the evolutionarily  ...  ., 2010; Kempe & Mesoudi, 2014 ) surprisingly few studies have examined the evolution of learning for social interaction dilemmas.  ... 
doi:10.1016/j.anbehav.2015.01.037 fatcat:cutxjqnj6zbfnagrery2vxqsb4

Emotional Multiagent Reinforcement Learning in Social Dilemmas [chapter]

Chao Yu, Minjie Zhang, Fenghui Ren
2013 Lecture Notes in Computer Science  
Without extra mechanisms or assumptions, directly applying multiagent reinforcement learning in social dilemmas will end up with convergence to the Nash equilibrium of mutual defection among the agents  ...  This paper investigates the importance of emotions in modifying agent learning behaviors in order to achieve cooperation in social dilemmas.  ...  reactions affect agent learning behaviors in social dilemmas.  ... 
doi:10.1007/978-3-642-44927-7_25 fatcat:aqqblwmrxvh77muj4ovpxfjh6m

Partner Selection for the Emergence of Cooperation in Multi-Agent Systems Using Reinforcement Learning [article]

Nicolas Anastassacos, Stephen Hailes, Mirco Musolesi
2019 arXiv   pre-print
Social dilemmas have been widely studied to explain how humans are able to cooperate in society.  ...  Our experiments reveal that agents trained with this dynamic learn a strategy that retaliates against defectors while promoting cooperation with other agents resulting in a prosocial society.  ...  Reinforcement Learning In Multi-Agent Social Dilemmas RL is a useful tool for understanding social dilemmas.  ... 
arXiv:1902.03185v4 fatcat:x2mocs7ofvhbniiwm5xasmaoaa

Testing the possibility to manage cooperation in CO2 crisis through mechanisms to face the dependence of the initial condition of trust using a simulation model

Jorge Andrick Parra-Valencia, Isaac Dyner-Rezonzew, María Cristina Serrano, Eliécer Pineda-Ballesteros, Adriana Rocío Lizcano-Dallos
2014 Revista UIS Ingenierías  
The simulation experiments offer support to our hypothesis about the possibility to manage cooperation in large-scale social dilemmas even if the trust's initial conditions are not enough to expect high  ...  We tested the possibility to promote and sustain cooperation through a combination of mechanisms using a simulation model in the CO2 crisis.  ...  The dynamic version of the Ostrom's mechanism of cooperation based on trust (Ostrom, 2000) for large-scale social dilemmas that we suggested worked under the conditions of this kind of social dilemmas  ... 
doaj:2bc663ec547b4d6c9417c58215d4c0b3 fatcat:zujvaqb73vhmpesf5k3broggvq

Autocurricula and the Emergence of Innovation from Social Interaction: A Manifesto for Multi-Agent Intelligence Research [article]

Joel Z. Leibo, Edward Hughes, Marc Lanctot, Thore Graepel
2019 arXiv   pre-print
The solution of one social task often begets new social tasks, continually generating novel challenges, and thereby promoting innovation.  ...  Here we explore the hypothesis that multi-agent systems sometimes display intrinsic dynamics arising from competition and cooperation that provide a naturally emergent curriculum, which we term an autocurriculum  ...  In addition, the first author would like to thank all the speakers, organizers, and attendees of the 2014 "Are there limits to evolution?"  ... 
arXiv:1903.00742v2 fatcat:xbx5ybhxkbhn7evdguxobyd2yi

Deep reinforcement learning models the emergent dynamics of human cooperation [article]

Kevin R. McKee, Edward Hughes, Tina O. Zhu, Martin J. Chadwick, Raphael Koster, Antonio Garcia Castaneda, Charlie Beattie, Thore Graepel, Matt Botvinick, Joel Z. Leibo
2021 arXiv   pre-print
spatial and temporal strategies for collective action in a social dilemma.  ...  We leverage multi-agent deep reinforcement learning to model how a social-cognitive mechanism--specifically, the intrinsic motivation to achieve a good reputation--steers group behavior toward specific  ...  dilemma for reinforcement learning agents.  ... 
arXiv:2103.04982v1 fatcat:wyj5zeflkjbo7lwqxc23cva5pi

Multi-Agent Reinforcement Learning and Human Social Factors in Climate Change Mitigation [article]

Kyle Tilbury, Jesse Hoey
2020 arXiv   pre-print
We propose applying multi-agent reinforcement learning (MARL) in this setting to develop intelligent agents that can influence the social factors at play in climate change mitigation.  ...  Climate change mitigation, a social dilemma made difficult by the inherent complexities of human behavior, has an impact at a global scale.  ...  Multi-agent reinforcement learning is commonly demonstrated with social dilemmas (Leibo et al. 2017; Peysakhovich and Lerer 2018; Tampuu et al. 2017; Jaques et al. 2019) .  ... 
arXiv:2002.05147v1 fatcat:mvafm3piezheldvksszmbe7nzm

Multi-agent Reinforcement Learning in Sequential Social Dilemmas [article]

Joel Z. Leibo, Vinicius Zambaldi, Marc Lanctot, Janusz Marecki, Thore Graepel
2017 arXiv   pre-print
We introduce sequential social dilemmas that share the mixed incentive structure of matrix game social dilemmas but also require agents to learn policies that implement their strategic intentions.  ...  In real-world social dilemmas these choices are temporally extended. Cooperativeness is a property that applies to policies, not elementary actions.  ...  Acknowledgments The authors would like to thank Chrisantha Fernando, Toby Ord, and Peter Sunehag for fruitful discussions in the leadup to this work, and Charles Beattie, Denis Teplyashin, and Stig Petersen  ... 
arXiv:1702.03037v1 fatcat:yejw2cbprracplaw3z5pg7ghiq

Online Learning in Iterated Prisoner's Dilemma to Mimic Human Behavior [article]

Baihan Lin, Djallel Bouneffouf, Guillermo Cecchi
2020 arXiv   pre-print
Results suggest that considering the current situation to make decision is the worst in this kind of social dilemma game.  ...  We propose to study online learning algorithm behavior in the Iterated Prisoner's Dilemma (IPD) game, where we explored the full spectrum of reinforcement learning agents: multi-armed bandits, contextual  ...  in this kind of social dilemma game.  ... 
arXiv:2006.06580v2 fatcat:qtujt27akbayxlvtplgg5g7ixe

Learning through Probing: a decentralized reinforcement learning architecture for social dilemmas [article]

Nicolas Anastassacos, Mirco Musolesi
2018 arXiv   pre-print
Multi-agent reinforcement learning has received significant interest in recent years notably due to the advancements made in deep reinforcement learning which have allowed for the developments of new architectures  ...  Using social dilemmas as the training ground, we present a novel learning architecture, Learning through Probing (LTP), where agents utilize a probing mechanism to incorporate how their opponent's behavior  ...  Figure 1 : 1 Payoff Matrix for Social Dilemmas and Iterated Prisoner's Dilemma.  ... 
arXiv:1809.10007v2 fatcat:z7uae6syqzhvjezuh4zhg44fam

Emotional Multiagent Reinforcement Learning in Spatial Social Dilemmas

Chao Yu, Minjie Zhang, Fenghui Ren, Guozhen Tan
2015 IEEE Transactions on Neural Networks and Learning Systems  
This paper investigates the possibility of exploiting emotions in agent learning in order to facilitate the emergence of cooperation in social dilemmas.  ...  Understanding how agents can achieve cooperation in social dilemmas through learning from local experience is a critical problem that has motivated researchers for decades.  ...  [37] used a combination of replicator dynamics and switching dynamics to model multiagent learning automata in multi-state Prisoner's Dilemma (PD) games.  ... 
doi:10.1109/tnnls.2015.2403394 pmid:25769173 fatcat:t5tyl2hqtbbs7ogcrjj5erjkpm
« Previous Showing results 1 — 15 out of 57,592 results