Matches in SemOpenAlex for { <https://semopenalex.org/work/W3097877554> ?p ?o ?g. }
- W3097877554 abstract "The paper provides a thorough investigation of Direct loss minimization (DLM), which optimizes the posterior to minimize predictive loss, in sparse Gaussian processes. For the conjugate case, we consider DLM for log-loss and DLM for square loss showing a significant performance improvement in both cases. The application of DLM in non-conjugate cases is more complex because the logarithm of expectation in the log-loss DLM objective is often intractable and simple sampling leads to biased estimates of gradients. The paper makes two technical contributions to address this. First, a new method using product sampling is proposed, which gives unbiased estimates of gradients (uPS) for the objective function. Second, a theoretical analysis of biased Monte Carlo estimates (bMC) shows that stochastic gradient descent converges despite the biased gradients. Experiments demonstrate empirical success of DLM. A comparison of the sampling methods shows that, while uPS is potentially more sample-efficient, bMC provides a better tradeoff in terms of convergence time and computational efficiency." @default.
- W3097877554 created "2020-11-09" @default.
- W3097877554 creator A5004165329 @default.
- W3097877554 creator A5013618541 @default.
- W3097877554 creator A5074409589 @default.
- W3097877554 date "2020-04-07" @default.
- W3097877554 modified "2023-09-27" @default.
- W3097877554 title "Direct loss minimization algorithms for sparse Gaussian processes" @default.
- W3097877554 cites W137285897 @default.
- W3097877554 cites W1533660737 @default.
- W3097877554 cites W1576452626 @default.
- W3097877554 cites W1746819321 @default.
- W3097877554 cites W1777124189 @default.
- W3097877554 cites W186151502 @default.
- W3097877554 cites W2036084078 @default.
- W3097877554 cites W2099119623 @default.
- W3097877554 cites W2099667085 @default.
- W3097877554 cites W2099768828 @default.
- W3097877554 cites W2186629860 @default.
- W3097877554 cites W2728803307 @default.
- W3097877554 cites W2752086744 @default.
- W3097877554 cites W2767137871 @default.
- W3097877554 cites W2770901349 @default.
- W3097877554 cites W2785967572 @default.
- W3097877554 cites W2890056239 @default.
- W3097877554 cites W2933774876 @default.
- W3097877554 cites W2962894765 @default.
- W3097877554 cites W2962897886 @default.
- W3097877554 cites W2963190151 @default.
- W3097877554 cites W2963566966 @default.
- W3097877554 cites W2964015990 @default.
- W3097877554 cites W2964126381 @default.
- W3097877554 cites W2970209593 @default.
- W3097877554 cites W2994724413 @default.
- W3097877554 cites W3014325088 @default.
- W3097877554 cites W3027291640 @default.
- W3097877554 cites W3034852262 @default.
- W3097877554 cites W3089766405 @default.
- W3097877554 cites W3106195168 @default.
- W3097877554 hasPublicationYear "2020" @default.
- W3097877554 type Work @default.
- W3097877554 sameAs 3097877554 @default.
- W3097877554 citedByCount "0" @default.
- W3097877554 crossrefType "posted-content" @default.
- W3097877554 hasAuthorship W3097877554A5004165329 @default.
- W3097877554 hasAuthorship W3097877554A5013618541 @default.
- W3097877554 hasAuthorship W3097877554A5074409589 @default.
- W3097877554 hasConcept C105795698 @default.
- W3097877554 hasConcept C106131492 @default.
- W3097877554 hasConcept C11413529 @default.
- W3097877554 hasConcept C121332964 @default.
- W3097877554 hasConcept C126255220 @default.
- W3097877554 hasConcept C134306372 @default.
- W3097877554 hasConcept C140779682 @default.
- W3097877554 hasConcept C147764199 @default.
- W3097877554 hasConcept C154945302 @default.
- W3097877554 hasConcept C162324750 @default.
- W3097877554 hasConcept C163716315 @default.
- W3097877554 hasConcept C19499675 @default.
- W3097877554 hasConcept C206688291 @default.
- W3097877554 hasConcept C2777303404 @default.
- W3097877554 hasConcept C31972630 @default.
- W3097877554 hasConcept C33923547 @default.
- W3097877554 hasConcept C39927690 @default.
- W3097877554 hasConcept C41008148 @default.
- W3097877554 hasConcept C50522688 @default.
- W3097877554 hasConcept C50644808 @default.
- W3097877554 hasConcept C52740198 @default.
- W3097877554 hasConcept C62520636 @default.
- W3097877554 hasConcept C81184566 @default.
- W3097877554 hasConceptScore W3097877554C105795698 @default.
- W3097877554 hasConceptScore W3097877554C106131492 @default.
- W3097877554 hasConceptScore W3097877554C11413529 @default.
- W3097877554 hasConceptScore W3097877554C121332964 @default.
- W3097877554 hasConceptScore W3097877554C126255220 @default.
- W3097877554 hasConceptScore W3097877554C134306372 @default.
- W3097877554 hasConceptScore W3097877554C140779682 @default.
- W3097877554 hasConceptScore W3097877554C147764199 @default.
- W3097877554 hasConceptScore W3097877554C154945302 @default.
- W3097877554 hasConceptScore W3097877554C162324750 @default.
- W3097877554 hasConceptScore W3097877554C163716315 @default.
- W3097877554 hasConceptScore W3097877554C19499675 @default.
- W3097877554 hasConceptScore W3097877554C206688291 @default.
- W3097877554 hasConceptScore W3097877554C2777303404 @default.
- W3097877554 hasConceptScore W3097877554C31972630 @default.
- W3097877554 hasConceptScore W3097877554C33923547 @default.
- W3097877554 hasConceptScore W3097877554C39927690 @default.
- W3097877554 hasConceptScore W3097877554C41008148 @default.
- W3097877554 hasConceptScore W3097877554C50522688 @default.
- W3097877554 hasConceptScore W3097877554C50644808 @default.
- W3097877554 hasConceptScore W3097877554C52740198 @default.
- W3097877554 hasConceptScore W3097877554C62520636 @default.
- W3097877554 hasConceptScore W3097877554C81184566 @default.
- W3097877554 hasLocation W30978775541 @default.
- W3097877554 hasOpenAccess W3097877554 @default.
- W3097877554 hasPrimaryLocation W30978775541 @default.
- W3097877554 hasRelatedWork W1971208478 @default.
- W3097877554 hasRelatedWork W2130314681 @default.
- W3097877554 hasRelatedWork W2145832734 @default.
- W3097877554 hasRelatedWork W2197142628 @default.