FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3199510319> ?p ?o ?g. }

• W3199510319 endingPage "655" @default.
• W3199510319 startingPage "638" @default.
• W3199510319 abstract "We investigate a new approach to compute the gradients of artificial neural networks (ANNs), based on the so-called push-out likelihood ratio method. Unlike the widely used backpropagation (BP) method that requires continuity of the loss function and the activation function, our approach bypasses this requirement by injecting artificial noises into the signals passed along the neurons. We show how this approach has a similar computational complexity as BP, and moreover is more advantageous in terms of removing the backward recursion and eliciting transparent formulas. We also formalize the connection between BP, a pivotal technique for training ANNs, and infinitesimal perturbation analysis, a classic path-wise derivative estimation approach, so that both our new proposed methods and BP can be better understood in the context of stochastic gradient estimation. Our approach allows efficient training for ANNs with more flexibility on the loss and activation functions, and shows empirical improvements on the robustness of ANNs under adversarial attacks and corruptions of natural noises. Summary of Contribution: Stochastic gradient estimation has been studied actively in simulation for decades and becomes more important in the era of machine learning and artificial intelligence. The stochastic gradient descent is a standard technique for training the artificial neural networks (ANNs), a pivotal problem in deep learning. The most popular stochastic gradient estimation technique is the backpropagation method. We find that the backpropagation method lies in the family of infinitesimal perturbation analysis, a path-wise gradient estimation technique in simulation. Moreover, we develop a new likelihood ratio-based method, another popular family of gradient estimation technique in simulation, for training more general ANNs, and demonstrate that the new training method can improve the robustness of the ANN." @default.
• W3199510319 created "2021-09-27" @default.
• W3199510319 creator A5005503619 @default.
• W3199510319 creator A5014068004 @default.
• W3199510319 creator A5027359861 @default.
• W3199510319 creator A5069167370 @default.
• W3199510319 creator A5071064725 @default.
• W3199510319 date "2022-01-01" @default.
• W3199510319 modified "2023-10-14" @default.
• W3199510319 title "A New Likelihood Ratio Method for Training Artificial Neural Networks" @default.
• W3199510319 cites W1498436455 @default.
• W3199510319 cites W1531253382 @default.
• W3199510319 cites W1825688215 @default.
• W3199510319 cites W1857539435 @default.
• W3199510319 cites W1965786092 @default.
• W3199510319 cites W1980287119 @default.
• W3199510319 cites W2009485448 @default.
• W3199510319 cites W2016207894 @default.
• W3199510319 cites W2026615512 @default.
• W3199510319 cites W2036347618 @default.
• W3199510319 cites W2046765929 @default.
• W3199510319 cites W2060228835 @default.
• W3199510319 cites W2092868272 @default.
• W3199510319 cites W2099220105 @default.
• W3199510319 cites W2118000334 @default.
• W3199510319 cites W2127255353 @default.
• W3199510319 cites W2130521152 @default.
• W3199510319 cites W2142087254 @default.
• W3199510319 cites W2149479912 @default.
• W3199510319 cites W2175202650 @default.
• W3199510319 cites W2311126677 @default.
• W3199510319 cites W2787618032 @default.
• W3199510319 cites W2963941964 @default.
• W3199510319 cites W4255949318 @default.
• W3199510319 doi "https://doi.org/10.1287/ijoc.2021.1088" @default.
• W3199510319 hasPublicationYear "2022" @default.
• W3199510319 type Work @default.
• W3199510319 sameAs 3199510319 @default.
• W3199510319 citedByCount "2" @default.
• W3199510319 countsByYear W31995103192023 @default.
• W3199510319 crossrefType "journal-article" @default.
• W3199510319 hasAuthorship W3199510319A5005503619 @default.
• W3199510319 hasAuthorship W3199510319A5014068004 @default.
• W3199510319 hasAuthorship W3199510319A5027359861 @default.
• W3199510319 hasAuthorship W3199510319A5069167370 @default.
• W3199510319 hasAuthorship W3199510319A5071064725 @default.
• W3199510319 hasBestOaLocation W31995103192 @default.
• W3199510319 hasConcept C104317684 @default.
• W3199510319 hasConcept C11413529 @default.
• W3199510319 hasConcept C115680565 @default.
• W3199510319 hasConcept C119857082 @default.
• W3199510319 hasConcept C126255220 @default.
• W3199510319 hasConcept C153258448 @default.
• W3199510319 hasConcept C154945302 @default.
• W3199510319 hasConcept C155032097 @default.
• W3199510319 hasConcept C185592680 @default.
• W3199510319 hasConcept C199360897 @default.
• W3199510319 hasConcept C206688291 @default.
• W3199510319 hasConcept C2777735758 @default.
• W3199510319 hasConcept C33923547 @default.
• W3199510319 hasConcept C38365724 @default.
• W3199510319 hasConcept C41008148 @default.
• W3199510319 hasConcept C50644808 @default.
• W3199510319 hasConcept C55493867 @default.
• W3199510319 hasConcept C63479239 @default.
• W3199510319 hasConceptScore W3199510319C104317684 @default.
• W3199510319 hasConceptScore W3199510319C11413529 @default.
• W3199510319 hasConceptScore W3199510319C115680565 @default.
• W3199510319 hasConceptScore W3199510319C119857082 @default.
• W3199510319 hasConceptScore W3199510319C126255220 @default.
• W3199510319 hasConceptScore W3199510319C153258448 @default.
• W3199510319 hasConceptScore W3199510319C154945302 @default.
• W3199510319 hasConceptScore W3199510319C155032097 @default.
• W3199510319 hasConceptScore W3199510319C185592680 @default.
• W3199510319 hasConceptScore W3199510319C199360897 @default.
• W3199510319 hasConceptScore W3199510319C206688291 @default.
• W3199510319 hasConceptScore W3199510319C2777735758 @default.
• W3199510319 hasConceptScore W3199510319C33923547 @default.
• W3199510319 hasConceptScore W3199510319C38365724 @default.
• W3199510319 hasConceptScore W3199510319C41008148 @default.
• W3199510319 hasConceptScore W3199510319C50644808 @default.
• W3199510319 hasConceptScore W3199510319C55493867 @default.
• W3199510319 hasConceptScore W3199510319C63479239 @default.
• W3199510319 hasIssue "1" @default.
• W3199510319 hasLocation W31995103191 @default.
• W3199510319 hasLocation W31995103192 @default.
• W3199510319 hasOpenAccess W3199510319 @default.
• W3199510319 hasPrimaryLocation W31995103191 @default.
• W3199510319 hasRelatedWork W1489449076 @default.
• W3199510319 hasRelatedWork W2118782552 @default.
• W3199510319 hasRelatedWork W2949203910 @default.
• W3199510319 hasRelatedWork W3159389381 @default.
• W3199510319 hasRelatedWork W3199510319 @default.
• W3199510319 hasRelatedWork W4206520803 @default.
• W3199510319 hasRelatedWork W4300155862 @default.
• W3199510319 hasRelatedWork W4362706668 @default.
• W3199510319 hasRelatedWork W4366280654 @default.
• W3199510319 hasRelatedWork W4385014862 @default.

• next