FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3071011332> ?p ?o ?g. }

• W3071011332 abstract "The generalized Gauss-Newton (GGN) approximation is often used to make practical Bayesian deep learning approaches scalable by replacing a second order derivative with a product of first order derivatives. In this paper we argue that the GGN approximation should be understood as a local linearization of the underlying Bayesian neural network (BNN), which turns the BNN into a generalized linear model (GLM). Because we use this linearized model for posterior inference, we should also predict using this modified model instead of the original one. We refer to this modified as GLM predictive and show that it effectively resolves common underfitting problems of the Laplace approximation. It extends previous results in this vein to general likelihoods and has an equivalent Gaussian process formulation, which enables alternative inference schemes for BNNs in function space. We demonstrate the effectiveness of our approach on several standard classification datasets as well as on out-of-distribution detection. We provide an implementation at this https URL." @default.
• W3071011332 created "2020-08-24" @default.
• W3071011332 creator A5016894844 @default.
• W3071011332 creator A5063386684 @default.
• W3071011332 creator A5066728592 @default.
• W3071011332 date "2020-08-19" @default.
• W3071011332 modified "2023-09-27" @default.
• W3071011332 title "Improving predictions of Bayesian neural nets via local linearization." @default.
• W3071011332 cites W137285897 @default.
• W3071011332 cites W1408639475 @default.
• W3071011332 cites W1533660737 @default.
• W3071011332 cites W1663973292 @default.
• W3071011332 cites W1746819321 @default.
• W3071011332 cites W1826234144 @default.
• W3071011332 cites W1970789124 @default.
• W3071011332 cites W2088538739 @default.
• W3071011332 cites W2107093743 @default.
• W3071011332 cites W2107386393 @default.
• W3071011332 cites W2138806252 @default.
• W3071011332 cites W2139701068 @default.
• W3071011332 cites W2164411961 @default.
• W3071011332 cites W2254249950 @default.
• W3071011332 cites W2478027467 @default.
• W3071011332 cites W2786857698 @default.
• W3071011332 cites W2804602357 @default.
• W3071011332 cites W2807776404 @default.
• W3071011332 cites W2809090039 @default.
• W3071011332 cites W2893995718 @default.
• W3071011332 cites W2899771611 @default.
• W3071011332 cites W2951266961 @default.
• W3071011332 cites W2955023002 @default.
• W3071011332 cites W2962739609 @default.
• W3071011332 cites W2962781217 @default.
• W3071011332 cites W2963433607 @default.
• W3071011332 cites W2963546708 @default.
• W3071011332 cites W2964121744 @default.
• W3071011332 cites W2964125128 @default.
• W3071011332 cites W2964319706 @default.
• W3071011332 cites W2965563166 @default.
• W3071011332 cites W2970111740 @default.
• W3071011332 cites W2970861023 @default.
• W3071011332 cites W2995638726 @default.
• W3071011332 cites W3004547119 @default.
• W3071011332 cites W3086499488 @default.
• W3071011332 cites W3100678069 @default.
• W3071011332 cites W3118608800 @default.
• W3071011332 cites W3146087985 @default.
• W3071011332 cites W3148198191 @default.
• W3071011332 cites W3157086450 @default.
• W3071011332 hasPublicationYear "2020" @default.
• W3071011332 type Work @default.
• W3071011332 sameAs 3071011332 @default.
• W3071011332 citedByCount "2" @default.
• W3071011332 countsByYear W30710113322021 @default.
• W3071011332 crossrefType "posted-content" @default.
• W3071011332 hasAuthorship W3071011332A5016894844 @default.
• W3071011332 hasAuthorship W3071011332A5063386684 @default.
• W3071011332 hasAuthorship W3071011332A5066728592 @default.
• W3071011332 hasConcept C107673813 @default.
• W3071011332 hasConcept C11210021 @default.
• W3071011332 hasConcept C11413529 @default.
• W3071011332 hasConcept C121332964 @default.
• W3071011332 hasConcept C126255220 @default.
• W3071011332 hasConcept C154945302 @default.
• W3071011332 hasConcept C158622935 @default.
• W3071011332 hasConcept C160234255 @default.
• W3071011332 hasConcept C163716315 @default.
• W3071011332 hasConcept C22243797 @default.
• W3071011332 hasConcept C2776135515 @default.
• W3071011332 hasConcept C2776214188 @default.
• W3071011332 hasConcept C2777472644 @default.
• W3071011332 hasConcept C28826006 @default.
• W3071011332 hasConcept C33923547 @default.
• W3071011332 hasConcept C41008148 @default.
• W3071011332 hasConcept C41587187 @default.
• W3071011332 hasConcept C50644808 @default.
• W3071011332 hasConcept C61326573 @default.
• W3071011332 hasConcept C62520636 @default.
• W3071011332 hasConceptScore W3071011332C107673813 @default.
• W3071011332 hasConceptScore W3071011332C11210021 @default.
• W3071011332 hasConceptScore W3071011332C11413529 @default.
• W3071011332 hasConceptScore W3071011332C121332964 @default.
• W3071011332 hasConceptScore W3071011332C126255220 @default.
• W3071011332 hasConceptScore W3071011332C154945302 @default.
• W3071011332 hasConceptScore W3071011332C158622935 @default.
• W3071011332 hasConceptScore W3071011332C160234255 @default.
• W3071011332 hasConceptScore W3071011332C163716315 @default.
• W3071011332 hasConceptScore W3071011332C22243797 @default.
• W3071011332 hasConceptScore W3071011332C2776135515 @default.
• W3071011332 hasConceptScore W3071011332C2776214188 @default.
• W3071011332 hasConceptScore W3071011332C2777472644 @default.
• W3071011332 hasConceptScore W3071011332C28826006 @default.
• W3071011332 hasConceptScore W3071011332C33923547 @default.
• W3071011332 hasConceptScore W3071011332C41008148 @default.
• W3071011332 hasConceptScore W3071011332C41587187 @default.
• W3071011332 hasConceptScore W3071011332C50644808 @default.
• W3071011332 hasConceptScore W3071011332C61326573 @default.
• W3071011332 hasConceptScore W3071011332C62520636 @default.
• W3071011332 hasLocation W30710113321 @default.
• W3071011332 hasOpenAccess W3071011332 @default.

• next