Filters








120,674 Hits in 6.7 sec

Planning with Abstract Learned Models While Learning Transferable Subtasks

John Winder, Stephanie Milani, Matthew Landen, Erebus Oh, Shane Parr, Shawn Squire, Marie DesJardins, Cynthia Matuszek
2020 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
We introduce an algorithm for model-based hierarchical reinforcement learning to acquire self-contained transition and reward models suitable for probabilistic planning at multiple levels of abstraction  ...  We call this framework Planning with Abstract Learned Models (PALM).  ...  IIS-1426452 and by DARPA under grants W911NF-15-1-0503 and D15AP00102.  ... 
doi:10.1609/aaai.v34i06.6555 fatcat:tclwwooz6repdfouimafrgbxwy

Planning with Abstract Learned Models While Learning Transferable Subtasks [article]

John Winder, Stephanie Milani, Matthew Landen, Erebus Oh, Shane Parr, Shawn Squire, Marie desJardins, Cynthia Matuszek
2020 arXiv   pre-print
We introduce an algorithm for model-based hierarchical reinforcement learning to acquire self-contained transition and reward models suitable for probabilistic planning at multiple levels of abstraction  ...  We call this framework Planning with Abstract Learned Models (PALM).  ...  IIS-1813223 and Grant No. IIS-1426452, and by DARPA under grants W911NF-15-1-0503 and D15AP00102.  ... 
arXiv:1912.07544v2 fatcat:m3fyipxdnnc7ppwmsio7wui22e

Constructing Abstraction Hierarchies Using a Skill-Symbol Loop

George Konidaris
2016 IJCAI International Joint Conference on Artificial Intelligence  
We describe how such a hierarchy can be used for fast planning, and illustrate the construction of an appropriate hierarchy for the Taxi domain.  ...  We describe a framework for building abstraction hierarchies whereby an agent alternates skill- and representation-construction phases to construct a sequence of increasingly abstract Markov decision processes  ...  Government is authorized to reproduce and distribute reprints for Governmental purposes notwithstanding any copyright notation thereon.  ... 
pmid:28579718 pmcid:PMC5455777 fatcat:3lmyu7qmbbgotkzxnw2wpwpe6q

Learning Abstract and Transferable Representations for Planning [article]

Steven James, Benjamin Rosman, George Konidaris
2022 arXiv   pre-print
We show how to combine these portable representations with problem-specific ones to generate a sound description of a specific task that can be used for abstract planning.  ...  We restrict our focus to learning representations for long-term planning, a class of problems that state-of-the-art learning methods are unable to solve.  ...  We intend to learn an abstract representation suitable for planning.  ... 
arXiv:2205.02092v1 fatcat:cbu5i3j7gbdbpiidircx5zdsu4

Constructing Abstraction Hierarchies Using a Skill-Symbol Loop [article]

George Konidaris
2015 arXiv   pre-print
We describe how such a hierarchy can be used for fast planning, and illustrate the construction of an appropriate hierarchy for the Taxi domain.  ...  We describe a framework for building abstraction hierarchies whereby an agent alternates skill- and representation-acquisition phases to construct a sequence of increasingly abstract Markov decision processes  ...  We used the above hierarchy to compute plans for three example queries, using dynamic programming and decision trees for planning and grounding classifiers, respectively.  ... 
arXiv:1509.07582v1 fatcat:he6cxbuyqrfbbn2md7yf6zhusi

Hierarchical case-based reasoning integrating case-based and decompositional problem-solving techniques for plant-control software design

B. Smyth, M.T. Keane, P. Cunningham
2001 IEEE Transactions on Knowledge and Data Engineering  
New problems are solved by retrieving and adapting the solutions to similar problems, solutions that have been stored and indexed for future reuse as cases in a case-base.  ...  The power of CBR is severely curtailed if problem solving is limited to the retrieval and adaptation of a single case, and for this reason the strategy of reusing multiple cases is immediately appealing  ...  In this way the solutions of Déjà Vu's abstract cases are similar to the abstract solution plans often used by the planning community.  ... 
doi:10.1109/69.956101 fatcat:mqvfjnf45vdsrmaicnho3axcmy

Partial Order Hierarchical Reinforcement Learning [chapter]

Bernhard Hengst
2008 Lecture Notes in Computer Science  
We go further and show how a problem can be automatically decomposed into a partial-order task-hierarchy, and solved using hierarchical reinforcement learning.  ...  In this paper the notion of a partial-order plan is extended to task-hierarchies. We introduce the concept of a partial-order taskhierarchy that decomposes a problem using multi-tasking actions.  ...  Acknowledgements NICTA is funded by the Australian Government as represented by the Department of Broadband, Communications and the Digital Economy and the Australian Research Council through the ICT Centre  ... 
doi:10.1007/978-3-540-89378-3_14 fatcat:bi6pxe4findg5oobpzqn4t6fha

Integrated planning representation using macros, abstractions, and cases

1997 Expert systems with applications  
The general operators in a successful plan derivation would be assessed for their potential usefulness, and some stored.  ...  Planning will be an essential part of future autonomous robots and integrated intelligent systems. This paper focuses on learning problem solving knowledge in planning systems.  ...  Conclusions The major contribution of this paper is the design of a learning system for a planner that combines macros, abstraction hierarchies, and case-based planning.  ... 
doi:10.1016/s0957-4174(97)83796-7 fatcat:gvxr3mliqrggzmlyylh56zdaee

Pedagogical plans as communication oriented objects

G. Olimpo, R.M. Bottino, J. Earp, M. Ott, F. Pozzi, M. Tavella
2010 Computers & Education  
The user too comprehends the plan in terms of a top-down process, where the specific steps of a learning activity are seen as originating from more general and abstract conceptualizations.  ...  Its purpose is to describe a structural model for pedagogical plans which can assist both authors and users.  ...  The upper nodes in the hierarchy typically provide an abstract and summarised description of a plan, which is especially useful for fostering comprehension and communication.  ... 
doi:10.1016/j.compedu.2010.02.011 fatcat:b3isrigeibhvliv2ttnp5flcum

Mining coverage statistics for websource selection in a mediator

Zaiqing Nie, Ullas Nambiar, Sreelakshmi Vaddi, Subbarao Kambhampati
2002 Proceedings of the eleventh international conference on Information and knowledge management - CIKM '02  
Despite this recognition there are no effective approaches for learning the needed statistics.  ...  Our approach uses a hierarchical classification of the queries, and threshold based variants of familiar data mining techniques to dynamically decide the level of resolution at which to learn the statistics  ...  From Figure 6 we can see that Ø × × is the next lowest abstract class for the query and hence the coverage statistics for Ø × × would be used to generate a plan.  ... 
doi:10.1145/584902.584917 fatcat:c5vf3n3hqjhbxpo3h3qnmqsngq

Mining coverage statistics for websource selection in a mediator

Zaiqing Nie, Ullas Nambiar, Sreelakshmi Vaddi, Subbarao Kambhampati
2002 Proceedings of the eleventh international conference on Information and knowledge management - CIKM '02  
Despite this recognition there are no effective approaches for learning the needed statistics.  ...  Our approach uses a hierarchical classification of the queries, and threshold based variants of familiar data mining techniques to dynamically decide the level of resolution at which to learn the statistics  ...  From Figure 6 we can see that Ø × × is the next lowest abstract class for the query and hence the coverage statistics for Ø × × would be used to generate a plan.  ... 
doi:10.1145/584792.584917 dblp:conf/cikm/NieNVK02 fatcat:5hovkw6jzjhrnd5upbxwe4deoq

Policy-contingent abstraction for robust robot control [article]

Joelle Pineau, Geoffrey Gordon, Sebastian Thrun
2012 arXiv   pre-print
Our approach is based on insights from the rich literature of hierarchical controllers and hierarchical MDPs.  ...  PolCA uses a human-designed task hierarchy which it traverses from the bottom up, learning an abstraction function and recursively-optimal policy for each subtask along the way.  ...  The Subsumption architecture for example uses a combination of hierarchi cal task partitioning and task-specific state abstraction to produce scalable control systems.  ... 
arXiv:1212.2495v1 fatcat:vtpbosxqdnb6rbof3loqtzezhm

Page 26 of Journal of Research and Practice in Information Technology Vol. 26, Issue 1 [page]

1994 Journal of Research and Practice in Information Technology  
It also describes an experimental methodology for use in measuring the effects of speed up learning Chapter 3, On Integrating Machine Learning with Planning by Gerald DeJong et al, and Chapter 4, The Role  ...  This decomposition is called an abstraction hierarchy, and finding abstraction hierarchies is the hard part Knoblock’s contribution in this volume is an automatic way to find abstraction hierarchies, based  ... 

A Unified Framework for Planning and Learning [chapter]

PAT LANGLEY, JOHN A. ALLEN
1993 Machine Learning Methods for Planning  
John Bresina, Mark Drummond, and Steve Minton also influenced our thinking. All of the above provided useful comments on an earlier draft.  ...  Acknowledgements We thank other members of the ICARUS group -Wayne Iba, Deepak Kulkarni, Kate McKusick, and Kevin Thompson for lengthy discussions that led to many of the ideas in this paper.  ...  Cases and Abstractions One common approach to encoding plan knowledge involves the use of abstract rules or schemas. For instance, Minton et al.'  ... 
doi:10.1016/b978-1-4832-0774-2.50015-9 fatcat:3x6fjaubxzfndbzc5pd4pugmhu

K-5 Teachers' Uses of Levels of Abstraction Focusing on Design

Jane Waite, Paul Curzon, William Marsh, Sue Sentance
2017 Proceedings of the 12th Workshop on Primary and Secondary Computing Education - WiPSCE '17  
Further exploration of levels of abstraction and particularly how the design level helps K-5 learners learn to program, in the same way that planning supports novices learning to write, is warranted.  ...  The teachers interviewed use the design level for both programming and writing.  ...  Teachers' understanding of the amount of detail needed for each level may impact on teaching and learning of levels of abstraction.  ... 
doi:10.1145/3137065.3137068 dblp:conf/wipsce/WaiteCMS17 fatcat:4mw5q6ebrvchhndk2pog77bsym
« Previous Showing results 1 — 15 out of 120,674 results