The Sharer's Dilemma in Collective Adaptive Systems of Self-Interested Agents
作者： Lenz Belzner , Martin Wirsing , Thomas Gabor , Kyrill Schmid , Thomy Phan
DOI:
关键词:
摘要: In collective adaptive systems (CAS), adaptation can be implemented by optimization wrt. utility. Agents in a CAS may self-interested, while their utilities depend on other agents' choices. Independent of agent yield poor individual and global reward due to locally interfering preferences. Joint scale poorly, is impossible if agents cannot expose preferences privacy or security issues. this paper, we study utility sharing for mitigating issue. Sharing with others incentivize individuals consider choices that are suboptimal but increase reward. We illustrate our approach variant distributed cross entropy optimization. Empirical results show increases expected payoff comparison without sharing. also investigate the effect greedy defectors sharing, self-interested agents. observe defection mean at expense individuals' payoff. empirically choice between yields fundamental dilemma CAS.
arxiv.org参考文章(20)
Wiebe Van der Hoek, Michael Wooldridge, Multi-Agent Systems Handbook of Knowledge Representation. pp. 887- 928 ,(2008)
Michael L. Littman, Markov games as a framework for multi-agent reinforcement learning Machine Learning Proceedings 1994. pp. 157- 163 ,(1994) , 10.1016/B978-1-55860-335-6.50027-1
Tuomas W. Sandholm, Robert H. Crites, Multiagent reinforcement learning in the Iterated Prisoner's Dilemma BioSystems. ,vol. 37, pp. 147- 166 ,(1996) , 10.1016/0303-2647(95)01551-5
Eric Sodomka, Michael Littman, Elizabeth Hilliard, Amy Greenwald, Coco-Q: Learning in Stochastic Games with Side Payments international conference on machine learning. pp. 1471- 1479 ,(2013)
Thomas C. Schelling, Hockey Helmets, Concealed Weapons, and Daylight Saving: A Study of Binary Choices With Externalities Journal of Conflict Resolution. ,vol. 17, pp. 381- 428 ,(1973) , 10.1177/002200277301700302
David Silver, Aja Huang, Chris J Maddison, Arthur Guez, Laurent Sifre, George Van Den Driessche, Julian Schrittwieser, Ioannis Antonoglou, Veda Panneershelvam, Marc Lanctot, Sander Dieleman, Dominik Grewe, John Nham, Nal Kalchbrenner, Ilya Sutskever, Timothy Lillicrap, Madeleine Leach, Koray Kavukcuoglu, Thore Graepel, Demis Hassabis, None, Mastering the game of Go with deep neural networks and tree search Nature. ,vol. 529, pp. 484- 489 ,(2016) , 10.1038/NATURE16961
Ferdinando Fioretto, Enrico Pontelli, William Yeoh, Distributed Constraint Optimization Problems and Applications: A Survey arXiv: Artificial Intelligence. ,(2016) , 10.1613/JAIR.5565
Shimon Whiteson, Nando de Freitas, Yannis M. Assael, Jakob N. Foerster, Learning to Communicate with Deep Multi-Agent Reinforcement Learning arXiv: Artificial Intelligence. ,(2016)
Thomas Anthony, David Barber, Zheng Tian, Thinking Fast and Slow with Deep Learning and Tree Search arXiv: Artificial Intelligence. ,(2017)
Alexander Peysakhovich, Adam Lerer, Maintaining cooperation in complex social dilemmas using deep reinforcement learning arXiv: Artificial Intelligence. ,(2017)