FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2894378449> ?p ?o ?g. }

• W2894378449 abstract "Neural approaches to sequence labeling often use a Conditional Random Field (CRF) to model their output dependencies, while Recurrent Neural Networks (RNN) are used for the same purpose in other tasks. We set out to establish RNNs as an attractive alternative to CRFs for sequence labeling. To do so, we address one of the RNN's most prominent shortcomings, the fact that it is not exposed to its own errors with the maximum-likelihood training. We frame the prediction of the output sequence as a sequential decision-making process, where we train the network with an adjusted actor-critic algorithm (AC-RNN). We comprehensively compare this strategy with maximum-likelihood training for both RNNs and CRFs on three structured-output tasks. The proposed AC-RNN efficiently matches the performance of the CRF on NER and CCG tagging, and outperforms it on Machine Transliteration. We also show that our training strategy is significantly better than other techniques for addressing RNN's exposure bias, such as Scheduled Sampling, and Self-Critical policy training." @default.
• W2894378449 created "2018-10-05" @default.
• W2894378449 creator A5008747488 @default.
• W2894378449 creator A5048302740 @default.
• W2894378449 creator A5074543563 @default.
• W2894378449 date "2018-09-30" @default.
• W2894378449 modified "2023-09-27" @default.
• W2894378449 title "Efficient Sequence Labeling with Actor-Critic Training" @default.
• W2894378449 cites W2009303086 @default.
• W2894378449 cites W2064675550 @default.
• W2894378449 cites W2074729706 @default.
• W2894378449 cites W2119717200 @default.
• W2894378449 cites W2144578941 @default.
• W2894378449 cites W2159636675 @default.
• W2894378449 cites W2250539671 @default.
• W2894378449 cites W2296283641 @default.
• W2894378449 cites W2461708311 @default.
• W2894378449 cites W2522606127 @default.
• W2894378449 cites W2545605631 @default.
• W2894378449 cites W2563900099 @default.
• W2894378449 cites W2776461464 @default.
• W2894378449 cites W2949335953 @default.
• W2894378449 cites W2949600092 @default.
• W2894378449 cites W2949888546 @default.
• W2894378449 cites W2950304420 @default.
• W2894378449 cites W2962902328 @default.
• W2894378449 cites W2962957031 @default.
• W2894378449 cites W2963084599 @default.
• W2894378449 cites W2963212250 @default.
• W2894378449 cites W2963248296 @default.
• W2894378449 cites W2963940534 @default.
• W2894378449 cites W2964121744 @default.
• W2894378449 cites W2964308564 @default.
• W2894378449 cites W2963768805 @default.
• W2894378449 hasPublicationYear "2018" @default.
• W2894378449 type Work @default.
• W2894378449 sameAs 2894378449 @default.
• W2894378449 citedByCount "1" @default.
• W2894378449 countsByYear W28943784492019 @default.
• W2894378449 crossrefType "posted-content" @default.
• W2894378449 hasAuthorship W2894378449A5008747488 @default.
• W2894378449 hasAuthorship W2894378449A5048302740 @default.
• W2894378449 hasAuthorship W2894378449A5074543563 @default.
• W2894378449 hasConcept C111919701 @default.
• W2894378449 hasConcept C119857082 @default.
• W2894378449 hasConcept C126042441 @default.
• W2894378449 hasConcept C127413603 @default.
• W2894378449 hasConcept C147168706 @default.
• W2894378449 hasConcept C152565575 @default.
• W2894378449 hasConcept C154945302 @default.
• W2894378449 hasConcept C177264268 @default.
• W2894378449 hasConcept C199360897 @default.
• W2894378449 hasConcept C201995342 @default.
• W2894378449 hasConcept C2775953691 @default.
• W2894378449 hasConcept C2778112365 @default.
• W2894378449 hasConcept C2780451532 @default.
• W2894378449 hasConcept C35639132 @default.
• W2894378449 hasConcept C41008148 @default.
• W2894378449 hasConcept C50644808 @default.
• W2894378449 hasConcept C54355233 @default.
• W2894378449 hasConcept C76155785 @default.
• W2894378449 hasConcept C86803240 @default.
• W2894378449 hasConcept C98045186 @default.
• W2894378449 hasConceptScore W2894378449C111919701 @default.
• W2894378449 hasConceptScore W2894378449C119857082 @default.
• W2894378449 hasConceptScore W2894378449C126042441 @default.
• W2894378449 hasConceptScore W2894378449C127413603 @default.
• W2894378449 hasConceptScore W2894378449C147168706 @default.
• W2894378449 hasConceptScore W2894378449C152565575 @default.
• W2894378449 hasConceptScore W2894378449C154945302 @default.
• W2894378449 hasConceptScore W2894378449C177264268 @default.
• W2894378449 hasConceptScore W2894378449C199360897 @default.
• W2894378449 hasConceptScore W2894378449C201995342 @default.
• W2894378449 hasConceptScore W2894378449C2775953691 @default.
• W2894378449 hasConceptScore W2894378449C2778112365 @default.
• W2894378449 hasConceptScore W2894378449C2780451532 @default.
• W2894378449 hasConceptScore W2894378449C35639132 @default.
• W2894378449 hasConceptScore W2894378449C41008148 @default.
• W2894378449 hasConceptScore W2894378449C50644808 @default.
• W2894378449 hasConceptScore W2894378449C54355233 @default.
• W2894378449 hasConceptScore W2894378449C76155785 @default.
• W2894378449 hasConceptScore W2894378449C86803240 @default.
• W2894378449 hasConceptScore W2894378449C98045186 @default.
• W2894378449 hasLocation W28943784491 @default.
• W2894378449 hasOpenAccess W2894378449 @default.
• W2894378449 hasPrimaryLocation W28943784491 @default.
• W2894378449 hasRelatedWork W2076440176 @default.
• W2894378449 hasRelatedWork W2112628236 @default.
• W2894378449 hasRelatedWork W2464072665 @default.
• W2894378449 hasRelatedWork W2552517936 @default.
• W2894378449 hasRelatedWork W2599829010 @default.
• W2894378449 hasRelatedWork W2603433898 @default.
• W2894378449 hasRelatedWork W2610308580 @default.
• W2894378449 hasRelatedWork W2747545581 @default.
• W2894378449 hasRelatedWork W2793805095 @default.
• W2894378449 hasRelatedWork W2931252866 @default.
• W2894378449 hasRelatedWork W2944915099 @default.
• W2894378449 hasRelatedWork W2951539240 @default.
• W2894378449 hasRelatedWork W2962824769 @default.
• W2894378449 hasRelatedWork W2964343439 @default.

• next