Matches in SemOpenAlex for { <https://semopenalex.org/work/W4287178397> ?p ?o ?g. }
Showing items 1 to 55 of
55
with 100 items per page.
- W4287178397 abstract "Bayesian inference allows to obtain useful information on the parameters of models, either in computational statistics or more recently in the context of Bayesian Neural Networks. The computational cost of usual Monte Carlo methods for sampling posterior laws in Bayesian inference scales linearly with the number of data points. One option to reduce it to a fraction of this cost is to resort to mini-batching in conjunction with unadjusted discretizations of Langevin dynamics, in which case only a random fraction of the data is used to estimate the gradient. However, this leads to an additional noise in the dynamics and hence a bias on the invariant measure which is sampled by the Markov chain. We advocate using the so-called Adaptive Langevin dynamics, which is a modification of standard inertial Langevin dynamics with a dynamical friction which automatically corrects for the increased noise arising from mini-batching. We investigate the practical relevance of the assumptions underpinning Adaptive Langevin (constant covariance for the estimation of the gradient, Gaussian minibatching noise), which are not satisfied in typical models of Bayesian inference, and quantify the bias induced by minibatching in this case. We also suggest a possible extension of AdL to further reduce the bias on the posterior distribution, by considering a dynamical friction depending on the current value of the parameter to sample." @default.
- W4287178397 created "2022-07-25" @default.
- W4287178397 creator A5015347592 @default.
- W4287178397 creator A5020480330 @default.
- W4287178397 date "2021-05-21" @default.
- W4287178397 modified "2023-10-16" @default.
- W4287178397 title "Quantifying the mini-batching error in Bayesian inference for Adaptive Langevin dynamics" @default.
- W4287178397 doi "https://doi.org/10.48550/arxiv.2105.10347" @default.
- W4287178397 hasPublicationYear "2021" @default.
- W4287178397 type Work @default.
- W4287178397 citedByCount "0" @default.
- W4287178397 crossrefType "posted-content" @default.
- W4287178397 hasAuthorship W4287178397A5015347592 @default.
- W4287178397 hasAuthorship W4287178397A5020480330 @default.
- W4287178397 hasBestOaLocation W42871783971 @default.
- W4287178397 hasConcept C107673813 @default.
- W4287178397 hasConcept C111350023 @default.
- W4287178397 hasConcept C11413529 @default.
- W4287178397 hasConcept C121332964 @default.
- W4287178397 hasConcept C121864883 @default.
- W4287178397 hasConcept C154945302 @default.
- W4287178397 hasConcept C160234255 @default.
- W4287178397 hasConcept C162376815 @default.
- W4287178397 hasConcept C2776214188 @default.
- W4287178397 hasConcept C2780004032 @default.
- W4287178397 hasConcept C41008148 @default.
- W4287178397 hasConcept C57830394 @default.
- W4287178397 hasConceptScore W4287178397C107673813 @default.
- W4287178397 hasConceptScore W4287178397C111350023 @default.
- W4287178397 hasConceptScore W4287178397C11413529 @default.
- W4287178397 hasConceptScore W4287178397C121332964 @default.
- W4287178397 hasConceptScore W4287178397C121864883 @default.
- W4287178397 hasConceptScore W4287178397C154945302 @default.
- W4287178397 hasConceptScore W4287178397C160234255 @default.
- W4287178397 hasConceptScore W4287178397C162376815 @default.
- W4287178397 hasConceptScore W4287178397C2776214188 @default.
- W4287178397 hasConceptScore W4287178397C2780004032 @default.
- W4287178397 hasConceptScore W4287178397C41008148 @default.
- W4287178397 hasConceptScore W4287178397C57830394 @default.
- W4287178397 hasLocation W42871783971 @default.
- W4287178397 hasOpenAccess W4287178397 @default.
- W4287178397 hasPrimaryLocation W42871783971 @default.
- W4287178397 hasRelatedWork W134473611 @default.
- W4287178397 hasRelatedWork W1972981664 @default.
- W4287178397 hasRelatedWork W2078368276 @default.
- W4287178397 hasRelatedWork W2099667085 @default.
- W4287178397 hasRelatedWork W2126586829 @default.
- W4287178397 hasRelatedWork W2290754432 @default.
- W4287178397 hasRelatedWork W2507814391 @default.
- W4287178397 hasRelatedWork W3166331719 @default.
- W4287178397 hasRelatedWork W4287178397 @default.
- W4287178397 hasRelatedWork W88131178 @default.
- W4287178397 isParatext "false" @default.
- W4287178397 isRetracted "false" @default.
- W4287178397 workType "article" @default.