Filters








1,443 Hits in 4.7 sec

Towards Open Ad Hoc Teamwork Using Graph-based Policy Learning [article]

Arrasy Rahman, Niklas Höpner, Filippos Christianos, Stefano V. Albrecht
2021 arXiv   pre-print
Ad hoc teamwork is the challenging problem of designing an autonomous agent which can adapt quickly to collaborate with teammates without prior coordination mechanisms, including joint training.  ...  Our solution builds on graph neural networks to learn agent models and joint-action value models under varying team compositions.  ...  Open ad hoc teamwork: This experiment simulates open ad hoc teamwork by allowing agents to enter and leave during episodes, both in training and testing.  ... 
arXiv:2006.10412v4 fatcat:ypvdqlbd6bhk7e32cemqvoribi

Making friends on the fly: Cooperating with new teammates

Samuel Barrett, Avi Rosenfeld, Sarit Kraus, Peter Stone
2017 Artificial Intelligence  
Recognizing that a key requirement of ad hoc teamwork is adaptability to previously unseen agents, the tests use more than 40 previously unknown teams on the first task and 7 previously unknown teams on  ...  This algorithm is instantiated in two forms: 1) PLASTIC-Model -which builds models of previous teammates' behaviors and plans behaviors online using these models and 2) PLASTIC-Policy -which learns policies  ...  Acknowledgments This work has taken place in the Learning Agents Research Group (LARG) at the Artificial Intelligence Laboratory, The University of Texas at Austin.  ... 
doi:10.1016/j.artint.2016.10.005 fatcat:ezxvbpjmszcrblqlnh6h6ye3xe

Ad hoc teamwork by learning teammates' task

Francisco S. Melo, Alberto Sardinha
2015 Autonomous Agents and Multi-Agent Systems  
This paper addresses the problem of ad hoc teamwork, where a learning agent engages in a cooperative task with other (unknown) agents.  ...  In our approach to the ad hoc teamwork problem, we represent tasks as fully cooperative matrix games.  ...  Acknowledgements The authors gratefully acknowledge the anonymous reviewers for the many useful suggestions that greatly improved the clarity of the presentation. This  ... 
doi:10.1007/s10458-015-9280-x fatcat:qhzwiqxncfas5am3pzfmdxw3iu

Teaching Social Behavior through Human Reinforcement for Ad hoc Teamwork -The STAR Framework [article]

Shani Alkoby, Avilash Rath, Peter Stone
2019 arXiv   pre-print
In this research, we introduce the STAR framework used to teach an ad hoc team of agents to act in accordance with human social norms.  ...  Most existing work considers the case of an individual agent attempting to learn a predefined set of rules.  ...  Ad hoc Teamwork The design of autonomous agents that can be a part of an ad hoc team is an important open problem in multiagent systems and as such has been widely studied [10, 31, 21] .  ... 
arXiv:1809.07880v3 fatcat:xxjmibb2wzfvtf2ir3crk3fi3m

Policy-Based Governance within Luna: Why We Developed Yet Another Agent Framework

Larry Bunch, Jeffrey M. Bradshaw, Tom Eskridge, Paul J. Feltovich, James Lott, Andrzej Uszok, Marco Carvalho
2012 2012 IEEE/WIC/ACM International Conferences on Web Intelligence and Intelligent Agent Technology  
In particular, we focus on how we use capabilities for comprehensive policy-based governance to ensure that key requirements for security, declarative specification of taskwork, and built-in support for  ...  KAoS, IHMC's ontology-based policy services framework, enables the semantically-rich and extensible semantics and the operational power and flexibility needed to realize these capabilities within Luna.  ...  : The observability features of agents can be used to support capabilities for policy learning-i.e., creating new KAoS policies programmatically based on patterns that are consistent and important to tasks  ... 
doi:10.1109/wi-iat.2012.272 dblp:conf/iat/BunchBEFLUC12 fatcat:ubkco3agyfdenm6pdettlwo6lm

Human-Agent Teamwork in Cyber Operations: Supporting Co-evolution of Tasks and Artifacts with Luna [chapter]

Larry Bunch, Jeffrey M. Bradshaw, Marco Carvalho, Tom Eskridge, Paul J. Feltovich, James Lott, Andrzej Uszok
2012 Lecture Notes in Computer Science  
In particular, we focus on how we use capabilities for comprehensive policy-based governance to ensure that key requirements for security, declarative specification of taskwork, and built-in support for  ...  NOC_A or NOC_B can create agents to perform ad hoc analysis and investigation tasks.  ...  : The observability features of agents can be used to support capabilities for policy learning-i.e., creating new KAoS policies programmatically based on patterns that are consistent and important to tasks  ... 
doi:10.1007/978-3-642-33690-4_7 fatcat:74ategn2ovdz7hkt23thngxdwa

Ad Hoc Teamwork With Behavior Switching Agents

Manish Ravula, Shani Alkoby, Peter Stone
2019 Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence  
This kind of cooperation, in which agents have to learn to cooperate on the fly, is called ad hoc teamwork.  ...  In this paper, we present a novel Convolutional-Neural-Network-based Change point Detection (CPD) algorithm for ad hoc teamwork.  ...  Acknowledgements This work has taken place in the Learning Agents Research Group (LARG) at UT Austin.  ... 
doi:10.24963/ijcai.2019/78 dblp:conf/ijcai/RavulaAS19 fatcat:krr7vjqo2nfcdnjdkuptct3baq

Three years of the RoboCup standard platform league drop-in player competition

Katie Genter, Tim Laue, Peter Stone
2016 Autonomous Agents and Multi-Agent Systems  
hoc teamwork using real robots.  ...  Specifically, in the following sections we discuss (1) what ad hoc teamwork researchers can learn from our experiences and (2) how the ad hoc teamwork strategies utilized in the Drop-in Player Competition  ... 
doi:10.1007/s10458-016-9353-5 fatcat:ktrd4y5owbcpxmyeze24tr3fvu

The Composition and Formation of Effective Teams. Computer Science meets Psychology [article]

Ewa Andrejczuk, Rita Berger, Juan A. Rodriguez-Aguilar, Carles Sierra, Víctor Marín-Puchades
2016 arXiv   pre-print
Agmon et al. (2014) consider ad-hoc settings with two types of agents: best-response agents and ad-hoc agents.  ...  On the other hand, ad-hoc agents have a more complete view of a team actions, agents' joint utilities and their action costs. Using such information, ad-hoc agents try to influence joint decisions.  ... 
arXiv:1610.08804v1 fatcat:vi2ysihpfra37otdhdqqydasru

DISTRIBUTED ON-LINE MULTI-AGENT OPTIMIZATION UNDER UNCERTAINTY: BALANCING EXPLORATION AND EXPLOITATION

MATTHEW E. TAYLOR, MANISH JAIN, PRATEEK TANDON, MAKOTO YOKOO, MILIND TAMBE
2011 Advances in Complex Systems  
to our expectations, we found that increasing teamwork in DCEE algorithms may lower team performance. In contrast, agents running DCOP algorithms improve their reward as teamwork increases.  ...  However, in many real-world multi-agent domains, a full model of the system is not known a priori and must instead be learned (e.g., any time the full model is unknown, such as when using ad-hoc teams  ...  While this approach enables agents to use unmodified single-agent RL algorithms, learning is likely to converge with poor performance, or fail to converge to a stable policy. (2) Agents collaborate to  ... 
doi:10.1142/s0219525911003104 fatcat:dafuu5c3xfbg3ma35qxhv4xuei

Scalable Evaluation of Multi-Agent Reinforcement Learning with Melting Pot [article]

Joel Z. Leibo, Edgar Duéñez-Guzmán, Alexander Sasha Vezhnevets, John P. Agapiou, Peter Sunehag, Raphael Koster, Jayd Matyas, Charles Beattie, Igor Mordatch, Thore Graepel
2021 arXiv   pre-print
Our contribution, Melting Pot, is a MARL evaluation suite that fills this gap, and uses reinforcement learning to reduce the human labor required to create novel test scenarios.  ...  Existing evaluation suites for multi-agent reinforcement learning (MARL) do not assess generalization to novel situations as their primary objective (unlike supervised-learning benchmarks).  ...  SC 4 Ad hoc teamwork with shaped A3C bots. This scenario tests ad-hoc teamwork.  ... 
arXiv:2107.06857v1 fatcat:ef7kcbzftnfy3bwchtccgmwvhy

Explainable Embodied Agents Through Social Cues

Sebastian Wallkötter, Silvia Tulli, Ginevra Castellano, Ana Paiva, Mohamed Chetouani
2021 ACM Transactions on Human-Robot Interaction (THRI)  
Another reason is that different authors use these terms in different ways.  ...  Additionally, we present a list of open questions and challenges that highlight areas that require further investigation by the community.  ...  Post hoc explainability mechanisms such as visualization, mapping the policy to natural language, or explanation are used to convert a non-explainable model into a more explainable one.  ... 
doi:10.1145/3457188 fatcat:2cbjfdac25eulbp2lpzuu6nuhq

Guest Editorial: Special Issue on High-Confidence City IoT for Collaborative Smart City Services

Dongxiao Yu, Xiuzhen Cheng, Falko Dressler, Dariusz R. Kowalski, Weifeng Lv
2020 IEEE Internet of Things Journal  
(CPABE) and key-policy attribute-based signature (KPABS).  ...  In "Batch-assisted verification scheme for Reducing Message Verification Delay of the Vehicular Ad Hoc Networks," Wu et al. proposed a message-batch-assisted verification scheme applied in areas with heavy  ... 
doi:10.1109/jiot.2020.3021495 dblp:journals/iotj/YuCDKL20 fatcat:qhvogwzf7rfgnea3ftj4eivs4m

Explainable Agents Through Social Cues: A Review [article]

Sebastian Wallkotter, Silvia Tulli, Ginevra Castellano, Ana Paiva, Mohamed Chetouani
2021 arXiv   pre-print
Another reason is that different authors use these terms in different ways.  ...  Additionally, we present a list of open questions and challenges that highlight areas that require further investigation by the community.  ...  Post-hoc explainability methods such as visualization, mapping the policy to natural language, or explanation are used to convert a non-explainable model into a more explainable one.  ... 
arXiv:2003.05251v3 fatcat:32o6n2bayzesfplybxbwxl5nle

Multi-Agent Curricula and Emergent Implicit Signaling [article]

Niko A. Grupen, Daniel D. Lee, Bart Selman
2022 arXiv   pre-print
Moreover, we examine the use of implicit signals in coordination through position-based social influence.  ...  Pursuit-evasion experiments show that our approach learns effective coordination, significantly outperforming sophisticated analytical and learned policies.  ...  In Figure 11 , the greedy pursuers form an ad-hoc triangular formation around the evader for a duration of 60 time-steps.  ... 
arXiv:2106.11156v3 fatcat:66zc76uz2vb5dhvbvorqz4wjjq
« Previous Showing results 1 — 15 out of 1,443 results