Matches in SemOpenAlex for { <https://semopenalex.org/work/W37272079> ?p ?o ?g. }
- W37272079 endingPage "330" @default.
- W37272079 startingPage "323" @default.
- W37272079 abstract "In stochastic planning problems formulated as factored Markov decision processes (MDPs), also called dynamic belief network MDPs (DBN-MDPs) (Boutilier, Dean, & Hanks 1999), finding the best policy (or conditional plan) is NP-hard. One of the difficulties comes from the fact that the number of conditionals required to specify the policy can grow to be exponential in the size of the representation for the MDP. Several recent algorithms have focused on finding an approximate policy by restricting the representation of conditionals using decision trees. We propose an alternative policy representation for Factored MDPs in terms of finite-state machine (FSM) controllers. Since practically speaking we are forced to limit the number of conditionals, we claim that there is a benefit to be had in using FSM controllers given that these controllers can use their internal state to maintain context information that might otherwise require a large conditional table or decision tree. Although the optimal policy might not be representable as a finite-state controller with a fixed amount of memory, we will be satisfied with finding a good policy; to that end, we derive a stochastic greedy-search algorithm based on recent developments in reinforcement learning (Baird & Moore 1999) and then demonstrate its performance in some example domains." @default.
- W37272079 created "2016-06-24" @default.
- W37272079 creator A5005123625 @default.
- W37272079 creator A5074569524 @default.
- W37272079 creator A5088683816 @default.
- W37272079 date "2000-04-14" @default.
- W37272079 modified "2023-09-23" @default.
- W37272079 title "Approximate solutions to factored Markov decision processes via greedy search in the space of finite state controllers" @default.
- W37272079 cites W1494689917 @default.
- W37272079 cites W1506485421 @default.
- W37272079 cites W1515851193 @default.
- W37272079 cites W1541084404 @default.
- W37272079 cites W1545148916 @default.
- W37272079 cites W1553182805 @default.
- W37272079 cites W1576452626 @default.
- W37272079 cites W1631187438 @default.
- W37272079 cites W1640774615 @default.
- W37272079 cites W1650504995 @default.
- W37272079 cites W180325379 @default.
- W37272079 cites W2061504687 @default.
- W37272079 cites W2107726111 @default.
- W37272079 cites W2119717200 @default.
- W37272079 cites W2121863487 @default.
- W37272079 cites W2164056559 @default.
- W37272079 cites W2168024904 @default.
- W37272079 cites W2334782222 @default.
- W37272079 cites W2912185451 @default.
- W37272079 cites W3023151133 @default.
- W37272079 cites W368409423 @default.
- W37272079 hasPublicationYear "2000" @default.
- W37272079 type Work @default.
- W37272079 sameAs 37272079 @default.
- W37272079 citedByCount "6" @default.
- W37272079 countsByYear W372720792012 @default.
- W37272079 crossrefType "proceedings-article" @default.
- W37272079 hasAuthorship W37272079A5005123625 @default.
- W37272079 hasAuthorship W37272079A5074569524 @default.
- W37272079 hasAuthorship W37272079A5088683816 @default.
- W37272079 hasConcept C105795698 @default.
- W37272079 hasConcept C106189395 @default.
- W37272079 hasConcept C119857082 @default.
- W37272079 hasConcept C126255220 @default.
- W37272079 hasConcept C151730666 @default.
- W37272079 hasConcept C154945302 @default.
- W37272079 hasConcept C159886148 @default.
- W37272079 hasConcept C163836022 @default.
- W37272079 hasConcept C17098449 @default.
- W37272079 hasConcept C17744445 @default.
- W37272079 hasConcept C188116033 @default.
- W37272079 hasConcept C199539241 @default.
- W37272079 hasConcept C2776359362 @default.
- W37272079 hasConcept C2779343474 @default.
- W37272079 hasConcept C33923547 @default.
- W37272079 hasConcept C41008148 @default.
- W37272079 hasConcept C72434380 @default.
- W37272079 hasConcept C86803240 @default.
- W37272079 hasConcept C94625758 @default.
- W37272079 hasConcept C97541855 @default.
- W37272079 hasConcept C98763669 @default.
- W37272079 hasConceptScore W37272079C105795698 @default.
- W37272079 hasConceptScore W37272079C106189395 @default.
- W37272079 hasConceptScore W37272079C119857082 @default.
- W37272079 hasConceptScore W37272079C126255220 @default.
- W37272079 hasConceptScore W37272079C151730666 @default.
- W37272079 hasConceptScore W37272079C154945302 @default.
- W37272079 hasConceptScore W37272079C159886148 @default.
- W37272079 hasConceptScore W37272079C163836022 @default.
- W37272079 hasConceptScore W37272079C17098449 @default.
- W37272079 hasConceptScore W37272079C17744445 @default.
- W37272079 hasConceptScore W37272079C188116033 @default.
- W37272079 hasConceptScore W37272079C199539241 @default.
- W37272079 hasConceptScore W37272079C2776359362 @default.
- W37272079 hasConceptScore W37272079C2779343474 @default.
- W37272079 hasConceptScore W37272079C33923547 @default.
- W37272079 hasConceptScore W37272079C41008148 @default.
- W37272079 hasConceptScore W37272079C72434380 @default.
- W37272079 hasConceptScore W37272079C86803240 @default.
- W37272079 hasConceptScore W37272079C94625758 @default.
- W37272079 hasConceptScore W37272079C97541855 @default.
- W37272079 hasConceptScore W37272079C98763669 @default.
- W37272079 hasOpenAccess W37272079 @default.
- W37272079 hasRelatedWork W1024197927 @default.
- W37272079 hasRelatedWork W153346180 @default.
- W37272079 hasRelatedWork W1546091409 @default.
- W37272079 hasRelatedWork W1548889916 @default.
- W37272079 hasRelatedWork W1575633987 @default.
- W37272079 hasRelatedWork W1650504995 @default.
- W37272079 hasRelatedWork W1688218840 @default.
- W37272079 hasRelatedWork W176737593 @default.
- W37272079 hasRelatedWork W2058735307 @default.
- W37272079 hasRelatedWork W2067718333 @default.
- W37272079 hasRelatedWork W2149896419 @default.
- W37272079 hasRelatedWork W2150821861 @default.
- W37272079 hasRelatedWork W2154623234 @default.
- W37272079 hasRelatedWork W2160067530 @default.
- W37272079 hasRelatedWork W2285658092 @default.
- W37272079 hasRelatedWork W2495570292 @default.
- W37272079 hasRelatedWork W2546801625 @default.