Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Filters








2,831 Hits in 3.6 sec

Multi-Agent Filtering with Infinitely Nested Beliefs

Luke S. Zettlemoyer, Brian Milch, Leslie Pack Kaelbling
2008 Neural Information Processing Systems  
In partially observable worlds with many agents, nested beliefs are formed when agents simultaneously reason about the unknown state of the world and the beliefs of the other agents.  ...  The multi-agent filtering problem is to efficiently represent and update these beliefs through time as the agents act in the world.  ...  We define infinitely nested beliefs by presenting an infinite sequence of finitely nested beliefs.  ... 
dblp:conf/nips/ZettlemoyerMK08 fatcat:z7eupkexrbbhrlz6xxgwl6zeku

Learning Others' Intentional Models in Multi-Agent Settings Using Interactive POMDPs

Yanlin Han, Piotr J. Gmytrasiewicz
2018 Neural Information Processing Systems  
It also effectively mitigates the belief space complexity due to the nested belief hierarchy.  ...  It extends POMDPs to multi-agent settings by including models of other agents in the state space and forming a hierarchical belief structure.  ...  The Interactive Particle Filter (I-PF) was devised as a filtering algorithm for interactive belief update in I-POMDP, which generalizes the classic particle filter algorithm to multi-agent settings [2  ... 
dblp:conf/nips/HanG18 fatcat:p4idc42sine5vgzablcehil55m

Multi-Agent Decentralized Belief Propagation on Graphs [article]

Yitao Chen, Deepanshu Vasal
2020 arXiv   pre-print
Our work appears to be the first study of decentralized belief propagation algorithm for networked multi-agent I-POMDPs.  ...  Within this setting, the collective goal of the agents is to maximize the globally averaged return over the network through exchanging information with their neighbors.  ...  The message type is crucial in reducing the complexity of I-POMDPs, for example, if the message is a belief, the agent does not have to maintain complex and infinitely nested beliefs of other agents.  ... 
arXiv:2011.04501v2 fatcat:xq5cwabluna7patoy5pq3izxuu

Monte Carlo Sampling Methods for Approximating Interactive POMDPs

P. Doshi, P. J Gmytrasiewicz
2009 The Journal of Artificial Intelligence Research  
physical world, about beliefs of other agents, and about their beliefs about others' beliefs.  ...  An extension of POMDPs to multiagent settings, called interactive POMDPs (I-POMDPs), replaces POMDP belief spaces with interactive hierarchical belief systems which represent an agent's belief about the  ...  Since the interactive beliefs may be infinitely nested, Gmytrasiewicz and Doshi (2005) defined finitely nested I-POMDPs as computable specializations of the infinitely nested ones.  ... 
doi:10.1613/jair.2630 fatcat:765tbcnhunanxad2jzdr72gtle

Interactive dynamic influence diagrams

Kyle Polich, Piotr Gmytrasiewicz
2007 Proceedings of the 6th international joint conference on Autonomous agents and multiagent systems - AAMAS '07  
I-POMDPs generalize POMDPs to multi-agent settings by including the models of other agents in the state space. In I-DIDs agents maintains their beliefs over models of other agents.  ...  This paper extends the framework of dynamic influence diagrams (DIDs) to the multi-agent setting.  ...  While the nesting of beliefs could be infinite, we assume finite nesting to ensure computability of the belief updates. 1 I-DIDs, analogously to DIDs, use a "forward" solution method, and do not rely on  ... 
doi:10.1145/1329125.1329166 dblp:conf/atal/PolichG07 fatcat:zpjhuo52sfalxdpe4b7jmxqtkm

Formal models and algorithms for decentralized decision making under uncertainty

Sven Seuken, Shlomo Zilberstein
2008 Autonomous Agents and Multi-Agent Systems  
This problem arises in many application domains, such as multi-robot coordination, manufacturing, information gathering, and load balancing.  ...  Such problems must be treated as decentralized decision problems because each agent may have different partial information about the other agents and about the state of the world.  ...  Finitely nested I-POMDPs Obviously, infinite nesting of beliefs in I-POMDPs leads to non-computable agent functions.  ... 
doi:10.1007/s10458-007-9026-5 fatcat:35cpuzfixvh6fecubllavumsjm

A Framework for Sequential Planning in Multi-Agent Settings

P. J. Gmytrasiewicz, P. Doshi
2005 The Journal of Artificial Intelligence Research  
This paper extends the framework of partially observable Markov decision processes (POMDPs) to multi-agent settings by incorporating the notion of agent models into the state space.  ...  Since the agent's beliefs may be arbitrarily nested, the optimal solutions to decision making problems are only asymptotically computable.  ...  Since the models of agents with infinitely nested beliefs correspond to agent functions which are not computable it is natural to consider finite nestings.  ... 
doi:10.1613/jair.1579 fatcat:arprt5mkmvgxxg5ei3qw2wgbwa

Interactive POMDP Lite: Towards Practical Planning to Predict and Exploit Intentions for Interacting with Self-Interested Agents [article]

Trong Nghia Hoang, Kian Hsiang Low
2013 arXiv   pre-print
A key challenge in non-cooperative multi-agent systems is that of developing efficient planning algorithms for intelligent agents to interact and perform effectively among boundedly rational, self-interested  ...  and acting optimally with respect to their predicted intentions.  ...  Notably, Interactive Particle Filtering (I-PF) [3] focused on alleviating the curse of dimensionality by generalizing the particle filtering technique to accommodate the multi-agent setting while Interactive  ... 
arXiv:1304.5159v1 fatcat:6vsbwvyiyvfvrbmuvlzcklf33e

Higher-Order Reasoning under Intent Uncertainty Reinforces the Hobbesian Trap (Appendix)

Otto Kuusela, Debraj Roy
2024 Zenodo  
In addition, the current belief is represented and updated separately from the search tree, for example with a particle filter.  ...  In an I-POMDP of agent 𝑗, other agents' decision making is typically modelled with nested I-POMDPs; each state has an associated I-POMDP model for every other agent.  ... 
doi:10.5281/zenodo.10631669 fatcat:bl4uvxstq5ewzb5dlnp3ojhec4

Forthcoming papers

1998 Artificial Intelligence  
We also provide several perspectives on memory-based reasoning from a multi-disciplinary point of view.  ...  In the process, we study several knowledge representation issues such as filtering, and restricted monotonicity with respect to NATs.  ...  Provetti, Formalizing narratives using nested circumscription Representing and reasoning about narratives together with the ability to do hypothetical reasoning is important for agents in a dynamic world  ... 
doi:10.1016/s0004-3702(98)90007-8 fatcat:kq2flquj35g4varimqe53txfym

Efficient Multi-agent Epistemic Planning: Teaching Planners About Nested Belief [article]

Christian Muise, Vaishak Belle, Paolo Felli, Sheila McIlraith, Tim Miller, Adrian R. Pearce, Liz Sonenberg
2021 arXiv   pre-print
We plan from the perspective of a single agent with the potential for goals and actions that involve nested beliefs, non-homogeneous agents, co-present observations, and the ability for one agent to reason  ...  We formally characterize our notion of planning with nested belief, and subsequently demonstrate how to automatically convert such problems into problems that appeal to classical planning technology for  ...  of complete knowledge over the nested belief of other agents.  ... 
arXiv:2110.02480v1 fatcat:w7r56nj45ba6dasbskdomrzwm4

Approximation Methods for Partially Observed Markov Decision Processes (POMDPs) [article]

Caleb M. Bowyer
2021 arXiv   pre-print
Then, I end the survey with some new research directions in .  ...  and developing guarantees for such multi-agent systems.  ...  The filter is used to update the agent or controller's belief-state as new information, specifically observations of the Markov chain as they are witnessed.  ... 
arXiv:2108.13965v1 fatcat:mnqm7mq7wnewbp6eg5kj2adpz4

Learning as Abductive Deliberations [chapter]

Budhitama Subagdja, Iyad Rahwan, Liz Sonenberg
2006 Lecture Notes in Computer Science  
With this model, the agent is capable of modifying its own plans on the run.  ...  We demonstrate that by abducing some complex structures of plan, the agent can also acquire complex structures of knowledge about its interaction with the environment.  ...  It works as an infinite loop of observe for updating belief, consider options for generating options, filter options for selecting intentions, select plan for selecting a plan instance, and intend that  ... 
doi:10.1007/978-3-540-36668-3_4 fatcat:d4j6vezqrzcopkvkadieats2ci

Small Infinitary Epistemic Logics

Tai-Wei Hu, Mamoru Kaneko, Nobu-Yuki Suzuki
2019 The Review of Symbolic Logic  
GL(L α ) has a sufficient expressive power to discuss intra/inter-personal beliefs with infinite lengths.  ...  , common beliefs, and infinite regress of beliefs.  ...  Our base logic is a finitary KD n with language L 0 (the set of formula); the agents have classical logical abilities and contradiction-free beliefs, described by the belief operators B i (·) for agents  ... 
doi:10.1017/s1755020319000029 fatcat:psa6c6fwrbbxxiluhbns5ysjbu

Incentive Decision Processes [article]

Sashank J. Reddi, Emma Brunskill
2012 arXiv   pre-print
We focus on the case where a principal interacts with a greedy agent whose preferences are hidden and static.  ...  We consider Incentive Decision Processes, where a principal seeks to reduce its costs due to another agent's behavior, by offering incentives to the agent for alternate behavior.  ...  Such representations can lead to infinite nestings of state estimates [11] .  ... 
arXiv:1210.4877v1 fatcat:doepj4jjszfmpkiwtd62dci5nu
« Previous Showing results 1 — 15 out of 2,831 results