3,198 Hits in 5.2 sec

Learning When to Quit: Meta-Reasoning for Motion Planning [article]

Yoonchang Sung, Leslie Pack Kaelbling, Tomás Lozano-Pérez
2021 arXiv   pre-print
We propose data-driven learning methods, model-based and model-free meta-reasoning, that are applicable to different environment distributions and agnostic to the choice of anytime motion planners.  ...  In this paper, we address the problem of deciding when to stop deliberation under bounded computational capacity, so called meta-reasoning, for anytime motion planning.  ...  Model-Free Meta-Reasoning In contrast with model-based meta-reasoning, learning a transition model is not required in the model-free approach.  ... 
arXiv:2103.04374v2 fatcat:ev7lape3dzh3zdqps3jtfk4cxq

Composing Real-Time Systems

Stuart J. Russell, Shlomo Zilberstein
1991 International Joint Conference on Artificial Intelligence  
We introduce a framework to measure the performance of anytime algorithms and solve the problem of constructing interruptible algorithms by a mathematical reduction to the problem of constructing contract  ...  We present a method to construct real-time systems using as components anytime algorithms whose quality of results degrades gracefully as computation time decreases.  ...  Our method is a meta-level approach in which the meta-level problem is limited to scheduling of anytime algorithms.  ... 
dblp:conf/ijcai/RussellZ91 fatcat:ke5vlquenzhsvmcy5vjcwbn4yi

Bounded Rationality in Multiagent Systems Using Decentralized Metareasoning [chapter]

Alan Carlin, Shlomo Zilberstein
2012 Intelligent Systems Reference Library  
In this paper, we analyze the problems that arise when several agents solve components of a larger problem, each using an anytime algorithm.  ...  Effective monitoring techniques have been developed to allow agents to stop their computation at the "right" time so as to optimize the overall time-dependent utility of the decision.  ...  One existing approach for meta-level coordination involves multiple agents that schedule a series of interrelated tasks [11] .  ... 
doi:10.1007/978-3-642-24647-0_1 fatcat:ukuds4kvgnguxfgxq4wjoysrxe

Coalition Formation among Bounded Rational Agents

Tuomas Sandholm, Victor R. Lesser
1995 International Joint Conference on Artificial Intelligence  
A model of bounded rationality is adopted, where computation resources are costly.  ...  A normative theory of coalitions among bounded rational (BR) agents is devised.  ...  It also allows us to sidestep the problem of having a meta-meta-level controlling the meta-level, a meta-meta-meta-level controlling the meta-meta-level, and so on ad infinitum.  ... 
dblp:conf/ijcai/SandholmL95 fatcat:xho4xwf5bfgz5b2bc6n535s2w4

Adaptive Rollout Length for Model-Based RL Using Model-Free Deep RL [article]

Abhinav Bhatia, Philip S. Thomas, Shlomo Zilberstein
2022 arXiv   pre-print
We use model-free deep reinforcement learning to solve the meta-level decision problem and demonstrate that our approach outperforms common heuristic baselines on two well-known reinforcement learning  ...  Model-based reinforcement learning promises to learn an optimal policy from fewer interactions with the environment compared to model-free reinforcement learning by learning an intermediate model of the  ...  been shown to be effective for solving decision-theoretic meta-level control problems in anytime planning (Bhatia et al., 2022) .  ... 
arXiv:2206.02380v2 fatcat:koisys57vrcvfiyvc636cvdlia

Automated algorithm configuration and design

Thomas Stützle, Manuel López-Ibáñez, Leslie Pérez-Cáceres
2022 Proceedings of the Genetic and Evolutionary Computation Conference Companion  
configuration apply powerful search techniques to design algorithms use computation power to explore algorithm design spaces free human creativity for higher level tasks T.  ...  Devise generalized LS (GLS) meta-metaheuristic 3 Describe possible GLS instantiations by a grammar 4 Convert the grammar to a parametric representation ✔ Allows use of standard automatic configuration  ... 
doi:10.1145/3520304.3533663 fatcat:wmge6w6i55c2joxl6y4w4extgi

Zero-Shot AutoML with Pretrained Models [article]

Ekrem Öztürk and Fabio Ferreira and Hadi S. Jomaa and Lars Schmidt-Thieme and Josif Grabocka and Frank Hutter
2022 arXiv   pre-print
To train this zero-shot model, we collect performance data for many DL pipelines on a large collection of datasets and meta-train on this data to minimize a pairwise ranking objective.  ...  Our domain-independent meta-learning approach learns a zero-shot surrogate model which, at test time, allows to select the right deep learning (DL) pipeline (including the pre-trained model and fine-tuning  ...  Wistuba et al. 2015 design a sequential model-free approach that optimizes the ranks of hyperparameter configurations based on their average performance over a collection of datasets.  ... 
arXiv:2206.08476v2 fatcat:wxqz2ppdlzborntmiqe4l2ojye

Discovering predictive ensembles for transfer learning and meta-learning

Pavel Kordík, Jan Černý, Tomáš Frýda
2017 Machine Learning  
Recent meta-learning approaches are oriented towards algorithm selection, optimization or recommendation of existing algorithms.  ...  Good-performing algorithms discovered by evolutionary algorithm can be reused on data sets of comparable complexity. Furthermore, these algorithms can be scaled up to model large data sets.  ...  Thanks to reviewers for their valuable time and comprehensive feedback.  ... 
doi:10.1007/s10994-017-5682-0 fatcat:xbxccvy55rbfdliw4prxm4t7cm

A framework for meta-level control in multi-agent systems

Anita Raja, Victor Lesser
2007 Autonomous Agents and Multi-Agent Systems  
The focus of this paper is how to make effective meta-level control decisions.  ...  The meta-level control approach that we present is based on the decision-theoretic use of an abstract representation of the agent state.  ...  Learning Algorithm.  ... 
doi:10.1007/s10458-006-9008-z fatcat:lfukqobaqzcephqkxvwzndf3l4

Metacognition in computation: A selected research review

Michael T. Cox
2005 Artificial Intelligence  
I discuss some of these aspects of cognition about cognition and the results concerning them from the point of view of the psychologist and the computer scientist, and I attempt to place them in the context  ...  I examine metacognition with respect to both problem solving (e.g., planning) and to comprehension (e.g., story understanding) processes of cognition.  2005 Published by Elsevier B.V.  ...  Monitoring informs the meta-level about the state of the object-level and thus allows the meta-level's model of the object level to be updated.  ... 
doi:10.1016/j.artint.2005.10.009 fatcat:6oghpyu5wrbe3djr6he4wvmx2y

A survey of techniques for achieving metadata interoperability

Bernhard Haslhofer, Wolfgang Klas
2010 ACM Computing Surveys  
Besides giving a general overview of the field of metadata interoperability, we provide a categorization of existing interoperability techniques, describe their characteristics, and compare their quality  ...  Achieving uniform access to media objects in heterogeneous media repositories requires dealing with the problem of metadata interoperability.  ...  ACKNOWLEDGMENTS The authors would like to thank Wolfgang Jochum and Bernhard Schandl as well as our referees for their valuable comments on this paper.  ... 
doi:10.1145/1667062.1667064 fatcat:otlqbh6zlrdwzootgfsmof55bi

Deliberation scheduling using GSMDPs in stochastic asynchronous domains

Kurt D. Krebsbach
2009 International Journal of Approximate Reasoning  
In this way, agents develop a continuous-time deliberation policy offline which can then be consulted to dynamically select deliberation-level and domain-level actions at plan execution-time.  ...  We propose a new decision-theoretic approach for solving execution-time deliberation scheduling problems using recent advances in Generalized Semi-Markov Decision Processes (GSMDPs).  ...  of this article.  ... 
doi:10.1016/j.ijar.2009.04.007 fatcat:japw4bafgbajton7srkabds5pm

Supporting Deliberative Real-Time AI Systems: A Fixed Priority Scheduling Approach

Yanching Chu, Alan Burns
2007 Real-Time Systems (ECRTS), Proceedings of the Euromicro Workshop on  
In this thesis it was shown that by employing a hybrid task model, for imprecise computation of the form "Prologue-Optional-Epilogue", a variety of AI algorithms with different hard and optional (anytime  ...  for allowing optional and anytime components to be executed for enhancing system utility.  ...  Since satisficing techniques seem to be a more natural approach than meta-reasoning, research has been focused on anytime algorithms and multiple approximate methods.  ... 
doi:10.1109/ecrts.2007.32 dblp:conf/ecrts/ChuB07 fatcat:ayfinirilbdxrhdjv5qc4wjyhu

Guiding Ebola Patients to Suitable Health Facilities: An SMS-based Approach [article]

Mohamad Trad, Raja Jurdak, Rajib Rana
2014 arXiv   pre-print
The added benefit of this approach is that it enables health care facilities to anticipate arrival of new potential Ebola cases.  ...  We propose to utilize mobile phone technology as a vehicle for people to report their symptoms and to receive immediate feedback about the health services readily available, and for predicting spatial  ...  It is particularly for this reason that we believe and advocate for the use of simple SMS messaging as a mechanism to guide Ebola patients to suitable health facilities.  ... 
arXiv:1410.3576v1 fatcat:vpslcfcbhraylf3hsa3ghwgooy

Optimal composition of real-time systems

Shlomo Zilberstein, Stuart Russell
1996 Artificial Intelligence  
Recent work by Dean, Horvitz and others has shown that anytime algorithms are a useful tool for real-time system design, since they allow computation time to be traded for decision quality.  ...  In order to construct complex systems, however, we need to be able to compose larger systems from smaller, reusable anytime modules.  ...  Tom Dean in particular inspired an initial interest in anytime algorithms and has continued to provide important insights and comments.  ... 
doi:10.1016/0004-3702(94)00074-3 fatcat:7jahnmsew5a7xf6r5jmenzocge
« Previous Showing results 1 — 15 out of 3,198 results