FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2807776404> ?p ?o ?g. }

• W2807776404 abstract "Uncertainty computation in deep learning is essential to design robust and reliable systems. Variational inference (VI) is a promising approach for such computation, but requires more effort to implement and execute compared to maximum-likelihood methods. In this paper, we propose new natural-gradient algorithms to reduce such efforts for Gaussian mean-field VI. Our algorithms can be implemented within the Adam optimizer by perturbing the network weights during gradient evaluations, and uncertainty estimates can be cheaply obtained by using the vector that adapts the learning rate. This requires lower memory, computation, and implementation effort than existing VI methods, while obtaining uncertainty estimates of comparable quality. Our empirical results confirm this and further suggest that the weight-perturbation in our algorithm could be useful for exploration in reinforcement learning and stochastic optimization." @default.
• W2807776404 created "2018-06-21" @default.
• W2807776404 creator A5019269745 @default.
• W2807776404 creator A5022716943 @default.
• W2807776404 creator A5029186201 @default.
• W2807776404 creator A5047246542 @default.
• W2807776404 creator A5053565482 @default.
• W2807776404 creator A5055819776 @default.
• W2807776404 date "2018-06-13" @default.
• W2807776404 modified "2023-10-02" @default.
• W2807776404 title "Fast and Scalable Bayesian Deep Learning by Weight-Perturbation in Adam" @default.
• W2807776404 cites W1486039312 @default.
• W2807776404 cites W1567512734 @default.
• W2807776404 cites W1909320841 @default.
• W2807776404 cites W2036084078 @default.
• W2807776404 cites W2104366350 @default.
• W2807776404 cites W2119717200 @default.
• W2807776404 cites W2127538960 @default.
• W2807776404 cites W2130984546 @default.
• W2807776404 cites W2138806252 @default.
• W2807776404 cites W2142916680 @default.
• W2807776404 cites W2146502635 @default.
• W2807776404 cites W2151965738 @default.
• W2807776404 cites W2152954918 @default.
• W2807776404 cites W2165150801 @default.
• W2807776404 cites W2210705382 @default.
• W2807776404 cites W2436219157 @default.
• W2807776404 cites W2485135680 @default.
• W2807776404 cites W2604395440 @default.
• W2807776404 cites W2619516334 @default.
• W2807776404 cites W2731982447 @default.
• W2807776404 cites W2753738274 @default.
• W2807776404 cites W2774412855 @default.
• W2807776404 cites W2786857698 @default.
• W2807776404 cites W2949568260 @default.
• W2807776404 cites W2950177356 @default.
• W2807776404 cites W2951266961 @default.
• W2807776404 cites W2951654389 @default.
• W2807776404 cites W2962915600 @default.
• W2807776404 cites W2963024489 @default.
• W2807776404 cites W2963173382 @default.
• W2807776404 cites W2963275203 @default.
• W2807776404 cites W2964027370 @default.
• W2807776404 cites W2964059111 @default.
• W2807776404 cites W2964121744 @default.
• W2807776404 cites W2964174623 @default.
• W2807776404 cites W3020853991 @default.
• W2807776404 cites W3140968660 @default.
• W2807776404 cites W71499226 @default.
• W2807776404 cites W759726671 @default.
• W2807776404 hasPublicationYear "2018" @default.
• W2807776404 type Work @default.
• W2807776404 sameAs 2807776404 @default.
• W2807776404 citedByCount "41" @default.
• W2807776404 countsByYear W28077764042017 @default.
• W2807776404 countsByYear W28077764042018 @default.
• W2807776404 countsByYear W28077764042019 @default.
• W2807776404 countsByYear W28077764042020 @default.
• W2807776404 countsByYear W28077764042021 @default.
• W2807776404 crossrefType "posted-content" @default.
• W2807776404 hasAuthorship W2807776404A5019269745 @default.
• W2807776404 hasAuthorship W2807776404A5022716943 @default.
• W2807776404 hasAuthorship W2807776404A5029186201 @default.
• W2807776404 hasAuthorship W2807776404A5047246542 @default.
• W2807776404 hasAuthorship W2807776404A5053565482 @default.
• W2807776404 hasAuthorship W2807776404A5055819776 @default.
• W2807776404 hasConcept C107673813 @default.
• W2807776404 hasConcept C108583219 @default.
• W2807776404 hasConcept C11413529 @default.
• W2807776404 hasConcept C119857082 @default.
• W2807776404 hasConcept C121332964 @default.
• W2807776404 hasConcept C126255220 @default.
• W2807776404 hasConcept C154945302 @default.
• W2807776404 hasConcept C160234255 @default.
• W2807776404 hasConcept C163716315 @default.
• W2807776404 hasConcept C2776214188 @default.
• W2807776404 hasConcept C2778049539 @default.
• W2807776404 hasConcept C2779377595 @default.
• W2807776404 hasConcept C33923547 @default.
• W2807776404 hasConcept C41008148 @default.
• W2807776404 hasConcept C45374587 @default.
• W2807776404 hasConcept C48044578 @default.
• W2807776404 hasConcept C62520636 @default.
• W2807776404 hasConcept C77088390 @default.
• W2807776404 hasConcept C97541855 @default.
• W2807776404 hasConceptScore W2807776404C107673813 @default.
• W2807776404 hasConceptScore W2807776404C108583219 @default.
• W2807776404 hasConceptScore W2807776404C11413529 @default.
• W2807776404 hasConceptScore W2807776404C119857082 @default.
• W2807776404 hasConceptScore W2807776404C121332964 @default.
• W2807776404 hasConceptScore W2807776404C126255220 @default.
• W2807776404 hasConceptScore W2807776404C154945302 @default.
• W2807776404 hasConceptScore W2807776404C160234255 @default.
• W2807776404 hasConceptScore W2807776404C163716315 @default.
• W2807776404 hasConceptScore W2807776404C2776214188 @default.
• W2807776404 hasConceptScore W2807776404C2778049539 @default.
• W2807776404 hasConceptScore W2807776404C2779377595 @default.
• W2807776404 hasConceptScore W2807776404C33923547 @default.
• W2807776404 hasConceptScore W2807776404C41008148 @default.
• W2807776404 hasConceptScore W2807776404C45374587 @default.

• next