Matches in SemOpenAlex for { <https://semopenalex.org/work/W2891044929> ?p ?o ?g. }
Showing items 1 to 92 of
92
with 100 items per page.
- W2891044929 endingPage "5747" @default.
- W2891044929 startingPage "5737" @default.
- W2891044929 abstract "Variational inference with α-divergences has been widely used in modern probabilistic machine learning. Compared to Kullback-Leibler (KL) divergence, a major advantage of using α-divergences (with positive α values) is their mass-covering property. However, estimating and optimizing α-divergences require to use importance sampling, which could have extremely large or infinite variances due to heavy tails of importance weights. In this paper, we propose a new class of tail-adaptive f-divergences that adaptively change the convex function f with the tail of the importance weights, in a way that theoretically guarantee finite moments, while simultaneously achieving mass-covering properties. We test our methods on Bayesian neural networks, as well as deep reinforcement learning in which our method is applied to improve a recent soft actor-critic (SAC) algorithm (Haarnoja et al., 2018). Our results show that our approach yields significant advantages compared with existing methods based on classical KL and α-divergences." @default.
- W2891044929 created "2018-09-27" @default.
- W2891044929 creator A5016340673 @default.
- W2891044929 creator A5057055806 @default.
- W2891044929 creator A5088664989 @default.
- W2891044929 date "2018-01-01" @default.
- W2891044929 modified "2023-09-27" @default.
- W2891044929 title "Variational Inference with Tail-adaptive f-Divergence" @default.
- W2891044929 hasPublicationYear "2018" @default.
- W2891044929 type Work @default.
- W2891044929 sameAs 2891044929 @default.
- W2891044929 citedByCount "17" @default.
- W2891044929 countsByYear W28910449292018 @default.
- W2891044929 countsByYear W28910449292019 @default.
- W2891044929 countsByYear W28910449292020 @default.
- W2891044929 countsByYear W28910449292021 @default.
- W2891044929 crossrefType "proceedings-article" @default.
- W2891044929 hasAuthorship W2891044929A5016340673 @default.
- W2891044929 hasAuthorship W2891044929A5057055806 @default.
- W2891044929 hasAuthorship W2891044929A5088664989 @default.
- W2891044929 hasConcept C107673813 @default.
- W2891044929 hasConcept C111472728 @default.
- W2891044929 hasConcept C112680207 @default.
- W2891044929 hasConcept C11413529 @default.
- W2891044929 hasConcept C126255220 @default.
- W2891044929 hasConcept C138885662 @default.
- W2891044929 hasConcept C145446738 @default.
- W2891044929 hasConcept C154945302 @default.
- W2891044929 hasConcept C160234255 @default.
- W2891044929 hasConcept C171752962 @default.
- W2891044929 hasConcept C189950617 @default.
- W2891044929 hasConcept C207390915 @default.
- W2891044929 hasConcept C2524010 @default.
- W2891044929 hasConcept C2776214188 @default.
- W2891044929 hasConcept C28826006 @default.
- W2891044929 hasConcept C33923547 @default.
- W2891044929 hasConcept C41008148 @default.
- W2891044929 hasConcept C41895202 @default.
- W2891044929 hasConcept C49937458 @default.
- W2891044929 hasConcept C50644808 @default.
- W2891044929 hasConcept C97541855 @default.
- W2891044929 hasConceptScore W2891044929C107673813 @default.
- W2891044929 hasConceptScore W2891044929C111472728 @default.
- W2891044929 hasConceptScore W2891044929C112680207 @default.
- W2891044929 hasConceptScore W2891044929C11413529 @default.
- W2891044929 hasConceptScore W2891044929C126255220 @default.
- W2891044929 hasConceptScore W2891044929C138885662 @default.
- W2891044929 hasConceptScore W2891044929C145446738 @default.
- W2891044929 hasConceptScore W2891044929C154945302 @default.
- W2891044929 hasConceptScore W2891044929C160234255 @default.
- W2891044929 hasConceptScore W2891044929C171752962 @default.
- W2891044929 hasConceptScore W2891044929C189950617 @default.
- W2891044929 hasConceptScore W2891044929C207390915 @default.
- W2891044929 hasConceptScore W2891044929C2524010 @default.
- W2891044929 hasConceptScore W2891044929C2776214188 @default.
- W2891044929 hasConceptScore W2891044929C28826006 @default.
- W2891044929 hasConceptScore W2891044929C33923547 @default.
- W2891044929 hasConceptScore W2891044929C41008148 @default.
- W2891044929 hasConceptScore W2891044929C41895202 @default.
- W2891044929 hasConceptScore W2891044929C49937458 @default.
- W2891044929 hasConceptScore W2891044929C50644808 @default.
- W2891044929 hasConceptScore W2891044929C97541855 @default.
- W2891044929 hasLocation W28910449291 @default.
- W2891044929 hasOpenAccess W2891044929 @default.
- W2891044929 hasPrimaryLocation W28910449291 @default.
- W2891044929 hasRelatedWork W1516111018 @default.
- W2891044929 hasRelatedWork W1530235965 @default.
- W2891044929 hasRelatedWork W1934021597 @default.
- W2891044929 hasRelatedWork W1959608418 @default.
- W2891044929 hasRelatedWork W2026653933 @default.
- W2891044929 hasRelatedWork W2115979064 @default.
- W2891044929 hasRelatedWork W2122882636 @default.
- W2891044929 hasRelatedWork W2150070703 @default.
- W2891044929 hasRelatedWork W2166851633 @default.
- W2891044929 hasRelatedWork W2397262406 @default.
- W2891044929 hasRelatedWork W2949743786 @default.
- W2891044929 hasRelatedWork W2963090522 @default.
- W2891044929 hasRelatedWork W2963173382 @default.
- W2891044929 hasRelatedWork W2963275229 @default.
- W2891044929 hasRelatedWork W2963383053 @default.
- W2891044929 hasRelatedWork W2963652898 @default.
- W2891044929 hasRelatedWork W2964318004 @default.
- W2891044929 hasRelatedWork W2964466098 @default.
- W2891044929 hasRelatedWork W3035059503 @default.
- W2891044929 hasRelatedWork W3101380508 @default.
- W2891044929 hasVolume "31" @default.
- W2891044929 isParatext "false" @default.
- W2891044929 isRetracted "false" @default.
- W2891044929 magId "2891044929" @default.
- W2891044929 workType "article" @default.