Matches in SemOpenAlex for { <https://semopenalex.org/work/W2952969080> ?p ?o ?g. }
- W2952969080 abstract "Hamiltonian Monte Carlo (HMC) is a widely deployed method to sample from high-dimensional distributions in Statistics and Machine learning. HMC is known to run very efficiently in practice and its popular second-order implementation has long been conjectured to run in $d^{1/4}$ gradient evaluations. Here we show that this conjecture is true when sampling from strongly log-concave target distributions that satisfy a weak third-order regularity property associated with the input data. Our regularity condition is weaker than the Lipschitz Hessian property and allows us to show faster convergence bounds for a much larger class of distributions than would be possible with the usual Lipschitz Hessian constant alone. Important distributions that satisfy our regularity condition include posterior distributions used in Bayesian logistic regression for which the data satisfies an incoherence property. Our result compares favorably with the best available bounds for the class of strongly log-concave distributions, which grow like $d^{{1}/{2}}$ gradient evaluations with the dimension. Moreover, our simulations on synthetic data suggest that, when our regularity condition is satisfied, leapfrog HMC performs better than its competitors -- both in terms of accuracy and in terms of the number of gradient evaluations it requires." @default.
- W2952969080 created "2019-06-27" @default.
- W2952969080 creator A5006627464 @default.
- W2952969080 creator A5063089732 @default.
- W2952969080 date "2018-02-24" @default.
- W2952969080 modified "2023-10-01" @default.
- W2952969080 title "Dimensionally Tight Bounds for Second-Order Hamiltonian Monte Carlo" @default.
- W2952969080 cites W1524688041 @default.
- W2952969080 cites W1566768190 @default.
- W2952969080 cites W1567512734 @default.
- W2952969080 cites W1809030594 @default.
- W2952969080 cites W1922512035 @default.
- W2952969080 cites W1981514681 @default.
- W2952969080 cites W1983452151 @default.
- W2952969080 cites W2005067744 @default.
- W2952969080 cites W2007902758 @default.
- W2952969080 cites W2014652454 @default.
- W2952969080 cites W2030911724 @default.
- W2952969080 cites W2032291279 @default.
- W2952969080 cites W2044410728 @default.
- W2952969080 cites W2053611995 @default.
- W2952969080 cites W2059448777 @default.
- W2952969080 cites W2072864444 @default.
- W2952969080 cites W2099878672 @default.
- W2952969080 cites W2100397110 @default.
- W2952969080 cites W2101687185 @default.
- W2952969080 cites W2132735979 @default.
- W2952969080 cites W2142274086 @default.
- W2952969080 cites W2144193737 @default.
- W2952969080 cites W2151717357 @default.
- W2952969080 cites W2166183437 @default.
- W2952969080 cites W2222154095 @default.
- W2952969080 cites W2272705390 @default.
- W2952969080 cites W2478027467 @default.
- W2952969080 cites W2562674776 @default.
- W2952969080 cites W2572702041 @default.
- W2952969080 cites W2577537660 @default.
- W2952969080 cites W2611204545 @default.
- W2952969080 cites W2746461338 @default.
- W2952969080 cites W2783417494 @default.
- W2952969080 cites W2787990337 @default.
- W2952969080 cites W2802698595 @default.
- W2952969080 cites W2883697110 @default.
- W2952969080 cites W2963183953 @default.
- W2952969080 cites W2963559717 @default.
- W2952969080 cites W2963977107 @default.
- W2952969080 cites W2964235356 @default.
- W2952969080 cites W2964301034 @default.
- W2952969080 cites W3123232560 @default.
- W2952969080 hasPublicationYear "2018" @default.
- W2952969080 type Work @default.
- W2952969080 sameAs 2952969080 @default.
- W2952969080 citedByCount "8" @default.
- W2952969080 countsByYear W29529690802017 @default.
- W2952969080 countsByYear W29529690802018 @default.
- W2952969080 countsByYear W29529690802019 @default.
- W2952969080 countsByYear W29529690802020 @default.
- W2952969080 countsByYear W29529690802021 @default.
- W2952969080 crossrefType "posted-content" @default.
- W2952969080 hasAuthorship W2952969080A5006627464 @default.
- W2952969080 hasAuthorship W2952969080A5063089732 @default.
- W2952969080 hasConcept C105795698 @default.
- W2952969080 hasConcept C111350023 @default.
- W2952969080 hasConcept C114614502 @default.
- W2952969080 hasConcept C126255220 @default.
- W2952969080 hasConcept C130787639 @default.
- W2952969080 hasConcept C13153151 @default.
- W2952969080 hasConcept C134306372 @default.
- W2952969080 hasConcept C19499675 @default.
- W2952969080 hasConcept C203616005 @default.
- W2952969080 hasConcept C22324862 @default.
- W2952969080 hasConcept C2780990831 @default.
- W2952969080 hasConcept C28826006 @default.
- W2952969080 hasConcept C33923547 @default.
- W2952969080 hasConceptScore W2952969080C105795698 @default.
- W2952969080 hasConceptScore W2952969080C111350023 @default.
- W2952969080 hasConceptScore W2952969080C114614502 @default.
- W2952969080 hasConceptScore W2952969080C126255220 @default.
- W2952969080 hasConceptScore W2952969080C130787639 @default.
- W2952969080 hasConceptScore W2952969080C13153151 @default.
- W2952969080 hasConceptScore W2952969080C134306372 @default.
- W2952969080 hasConceptScore W2952969080C19499675 @default.
- W2952969080 hasConceptScore W2952969080C203616005 @default.
- W2952969080 hasConceptScore W2952969080C22324862 @default.
- W2952969080 hasConceptScore W2952969080C2780990831 @default.
- W2952969080 hasConceptScore W2952969080C28826006 @default.
- W2952969080 hasConceptScore W2952969080C33923547 @default.
- W2952969080 hasLocation W29529690801 @default.
- W2952969080 hasOpenAccess W2952969080 @default.
- W2952969080 hasPrimaryLocation W29529690801 @default.
- W2952969080 hasRelatedWork W1995772505 @default.
- W2952969080 hasRelatedWork W2036346468 @default.
- W2952969080 hasRelatedWork W2222154095 @default.
- W2952969080 hasRelatedWork W2338000073 @default.
- W2952969080 hasRelatedWork W2345444914 @default.
- W2952969080 hasRelatedWork W2475426136 @default.
- W2952969080 hasRelatedWork W2562674776 @default.
- W2952969080 hasRelatedWork W2746461338 @default.
- W2952969080 hasRelatedWork W2788368350 @default.
- W2952969080 hasRelatedWork W2802698595 @default.