Matches in SemOpenAlex for { <https://semopenalex.org/work/W3035527463> ?p ?o ?g. }
Showing items 1 to 92 of
92
with 100 items per page.
- W3035527463 endingPage "3667" @default.
- W3035527463 startingPage "3658" @default.
- W3035527463 abstract "Off-policy evaluation in reinforcement learning offers the chance of using observational data to improve future outcomes in domains such as healthcare and education, but safe deployment in high stakes settings requires ways of assessing its validity. Traditional measures such as confidence intervals may be insufficient due to noise, limited data and confounding. In this paper we develop a method that could serve as a hybrid human-AI system, to enable human experts to analyze the validity of policy evaluation estimates. This is accomplished by highlighting observations in the data whose removal will have a large effect on the OPE estimate, and formulating a set of rules for choosing which ones to present to domain experts for validation. We develop methods to compute exactly the influence functions for fitted Q-evaluation with two different function classes: kernel-based and linear least squares, as well as importance sampling methods. Experiments on medical simulations and real-world intensive care unit data demonstrate that our method can be used to identify limitations in the evaluation process and make evaluation more robust." @default.
- W3035527463 created "2020-06-19" @default.
- W3035527463 creator A5025828990 @default.
- W3035527463 creator A5031401755 @default.
- W3035527463 creator A5038771285 @default.
- W3035527463 creator A5045518550 @default.
- W3035527463 creator A5064116762 @default.
- W3035527463 creator A5079293522 @default.
- W3035527463 creator A5084989076 @default.
- W3035527463 date "2020-07-12" @default.
- W3035527463 modified "2023-09-26" @default.
- W3035527463 title "Interpretable Off-Policy Evaluation in Reinforcement Learning by Highlighting Influential Transitions" @default.
- W3035527463 hasPublicationYear "2020" @default.
- W3035527463 type Work @default.
- W3035527463 sameAs 3035527463 @default.
- W3035527463 citedByCount "5" @default.
- W3035527463 countsByYear W30355274632020 @default.
- W3035527463 countsByYear W30355274632021 @default.
- W3035527463 crossrefType "proceedings-article" @default.
- W3035527463 hasAuthorship W3035527463A5025828990 @default.
- W3035527463 hasAuthorship W3035527463A5031401755 @default.
- W3035527463 hasAuthorship W3035527463A5038771285 @default.
- W3035527463 hasAuthorship W3035527463A5045518550 @default.
- W3035527463 hasAuthorship W3035527463A5064116762 @default.
- W3035527463 hasAuthorship W3035527463A5079293522 @default.
- W3035527463 hasAuthorship W3035527463A5084989076 @default.
- W3035527463 hasConcept C105002631 @default.
- W3035527463 hasConcept C105339364 @default.
- W3035527463 hasConcept C105795698 @default.
- W3035527463 hasConcept C111919701 @default.
- W3035527463 hasConcept C114614502 @default.
- W3035527463 hasConcept C115961682 @default.
- W3035527463 hasConcept C119857082 @default.
- W3035527463 hasConcept C124101348 @default.
- W3035527463 hasConcept C154945302 @default.
- W3035527463 hasConcept C177264268 @default.
- W3035527463 hasConcept C199360897 @default.
- W3035527463 hasConcept C23131810 @default.
- W3035527463 hasConcept C33923547 @default.
- W3035527463 hasConcept C41008148 @default.
- W3035527463 hasConcept C58328972 @default.
- W3035527463 hasConcept C74193536 @default.
- W3035527463 hasConcept C97541855 @default.
- W3035527463 hasConcept C98045186 @default.
- W3035527463 hasConcept C99498987 @default.
- W3035527463 hasConceptScore W3035527463C105002631 @default.
- W3035527463 hasConceptScore W3035527463C105339364 @default.
- W3035527463 hasConceptScore W3035527463C105795698 @default.
- W3035527463 hasConceptScore W3035527463C111919701 @default.
- W3035527463 hasConceptScore W3035527463C114614502 @default.
- W3035527463 hasConceptScore W3035527463C115961682 @default.
- W3035527463 hasConceptScore W3035527463C119857082 @default.
- W3035527463 hasConceptScore W3035527463C124101348 @default.
- W3035527463 hasConceptScore W3035527463C154945302 @default.
- W3035527463 hasConceptScore W3035527463C177264268 @default.
- W3035527463 hasConceptScore W3035527463C199360897 @default.
- W3035527463 hasConceptScore W3035527463C23131810 @default.
- W3035527463 hasConceptScore W3035527463C33923547 @default.
- W3035527463 hasConceptScore W3035527463C41008148 @default.
- W3035527463 hasConceptScore W3035527463C58328972 @default.
- W3035527463 hasConceptScore W3035527463C74193536 @default.
- W3035527463 hasConceptScore W3035527463C97541855 @default.
- W3035527463 hasConceptScore W3035527463C98045186 @default.
- W3035527463 hasConceptScore W3035527463C99498987 @default.
- W3035527463 hasOpenAccess W3035527463 @default.
- W3035527463 hasRelatedWork W2033025724 @default.
- W3035527463 hasRelatedWork W2121863487 @default.
- W3035527463 hasRelatedWork W2132836414 @default.
- W3035527463 hasRelatedWork W2183766570 @default.
- W3035527463 hasRelatedWork W2401474588 @default.
- W3035527463 hasRelatedWork W2531520581 @default.
- W3035527463 hasRelatedWork W2558073743 @default.
- W3035527463 hasRelatedWork W2576718746 @default.
- W3035527463 hasRelatedWork W2623906388 @default.
- W3035527463 hasRelatedWork W2955053048 @default.
- W3035527463 hasRelatedWork W2964251854 @default.
- W3035527463 hasRelatedWork W2996737117 @default.
- W3035527463 hasRelatedWork W2999930990 @default.
- W3035527463 hasRelatedWork W3034692406 @default.
- W3035527463 hasRelatedWork W3036121832 @default.
- W3035527463 hasRelatedWork W3089192979 @default.
- W3035527463 hasRelatedWork W3124056174 @default.
- W3035527463 hasRelatedWork W3139330167 @default.
- W3035527463 hasRelatedWork W3208738391 @default.
- W3035527463 hasRelatedWork W69978452 @default.
- W3035527463 hasVolume "1" @default.
- W3035527463 isParatext "false" @default.
- W3035527463 isRetracted "false" @default.
- W3035527463 magId "3035527463" @default.
- W3035527463 workType "article" @default.