Matches in SemOpenAlex for { <https://semopenalex.org/work/W2891958946> ?p ?o ?g. }
Showing items 1 to 95 of
95
with 100 items per page.
- W2891958946 abstract "Maximum-likelihood estimation (MLE) is one of the most widely used approaches for training structured prediction models for text-generation based natural language processing applications. However, besides exposure bias, models trained with MLE suffer from wrong objective problem where they are trained to maximize the word-level correct next step prediction, but are evaluated with respect to sequence-level discrete metrics such as ROUGE and BLEU. Several variants of policy-gradient methods address some of these problems by optimizing for final discrete evaluation metrics and showing improvements over MLE training for downstream tasks like text summarization and machine translation. However, policy-gradient methods suffers from high sample variance, making the training process very difficult and unstable. In this paper, we present an alternative direction towards mitigating this problem by introducing a new objective (CaLcs) based on a differentiable surrogate of longest common subsequence (LCS) measure that captures sequence-level structure similarity. Experimental results on abstractive summarization and machine translation validate the effectiveness of the proposed approach." @default.
- W2891958946 created "2018-09-27" @default.
- W2891958946 creator A5010253402 @default.
- W2891958946 creator A5027763497 @default.
- W2891958946 creator A5048500970 @default.
- W2891958946 creator A5083465496 @default.
- W2891958946 date "2018-01-01" @default.
- W2891958946 modified "2023-10-18" @default.
- W2891958946 title "CaLcs: Continuously Approximating Longest Common Subsequence for Sequence Level Optimization" @default.
- W2891958946 cites W1544827683 @default.
- W2891958946 cites W1902237438 @default.
- W2891958946 cites W2081265723 @default.
- W2891958946 cites W2101105183 @default.
- W2891958946 cites W2108325777 @default.
- W2891958946 cites W2130942839 @default.
- W2891958946 cites W2153653739 @default.
- W2891958946 cites W2157331557 @default.
- W2891958946 cites W2163548102 @default.
- W2891958946 cites W2307381258 @default.
- W2891958946 cites W2311921240 @default.
- W2891958946 cites W2467173223 @default.
- W2891958946 cites W2507756961 @default.
- W2891958946 cites W2525778437 @default.
- W2891958946 cites W2574535369 @default.
- W2891958946 cites W2606974598 @default.
- W2891958946 cites W2612675303 @default.
- W2891958946 cites W2741375528 @default.
- W2891958946 cites W2767989436 @default.
- W2891958946 cites W2772943665 @default.
- W2891958946 cites W2962965405 @default.
- W2891958946 cites W2963120839 @default.
- W2891958946 cites W2963125472 @default.
- W2891958946 cites W2963248296 @default.
- W2891958946 cites W2963403868 @default.
- W2891958946 cites W2963463964 @default.
- W2891958946 cites W2963929190 @default.
- W2891958946 cites W2963962369 @default.
- W2891958946 cites W2964121744 @default.
- W2891958946 cites W2964308564 @default.
- W2891958946 cites W2964352247 @default.
- W2891958946 cites W648786980 @default.
- W2891958946 cites W2963768805 @default.
- W2891958946 doi "https://doi.org/10.18653/v1/d18-1406" @default.
- W2891958946 hasPublicationYear "2018" @default.
- W2891958946 type Work @default.
- W2891958946 sameAs 2891958946 @default.
- W2891958946 citedByCount "5" @default.
- W2891958946 countsByYear W28919589462019 @default.
- W2891958946 countsByYear W28919589462022 @default.
- W2891958946 countsByYear W28919589462023 @default.
- W2891958946 crossrefType "proceedings-article" @default.
- W2891958946 hasAuthorship W2891958946A5010253402 @default.
- W2891958946 hasAuthorship W2891958946A5027763497 @default.
- W2891958946 hasAuthorship W2891958946A5048500970 @default.
- W2891958946 hasAuthorship W2891958946A5083465496 @default.
- W2891958946 hasBestOaLocation W28919589461 @default.
- W2891958946 hasConcept C11413529 @default.
- W2891958946 hasConcept C120098539 @default.
- W2891958946 hasConcept C134306372 @default.
- W2891958946 hasConcept C137877099 @default.
- W2891958946 hasConcept C170006305 @default.
- W2891958946 hasConcept C2778112365 @default.
- W2891958946 hasConcept C33923547 @default.
- W2891958946 hasConcept C34388435 @default.
- W2891958946 hasConcept C41008148 @default.
- W2891958946 hasConcept C54355233 @default.
- W2891958946 hasConcept C86803240 @default.
- W2891958946 hasConceptScore W2891958946C11413529 @default.
- W2891958946 hasConceptScore W2891958946C120098539 @default.
- W2891958946 hasConceptScore W2891958946C134306372 @default.
- W2891958946 hasConceptScore W2891958946C137877099 @default.
- W2891958946 hasConceptScore W2891958946C170006305 @default.
- W2891958946 hasConceptScore W2891958946C2778112365 @default.
- W2891958946 hasConceptScore W2891958946C33923547 @default.
- W2891958946 hasConceptScore W2891958946C34388435 @default.
- W2891958946 hasConceptScore W2891958946C41008148 @default.
- W2891958946 hasConceptScore W2891958946C54355233 @default.
- W2891958946 hasConceptScore W2891958946C86803240 @default.
- W2891958946 hasLocation W28919589461 @default.
- W2891958946 hasOpenAccess W2891958946 @default.
- W2891958946 hasPrimaryLocation W28919589461 @default.
- W2891958946 hasRelatedWork W1533815075 @default.
- W2891958946 hasRelatedWork W1974063126 @default.
- W2891958946 hasRelatedWork W2035123411 @default.
- W2891958946 hasRelatedWork W2077484872 @default.
- W2891958946 hasRelatedWork W2079700310 @default.
- W2891958946 hasRelatedWork W2131191265 @default.
- W2891958946 hasRelatedWork W2289394377 @default.
- W2891958946 hasRelatedWork W2375943101 @default.
- W2891958946 hasRelatedWork W2565438332 @default.
- W2891958946 hasRelatedWork W776417157 @default.
- W2891958946 isParatext "false" @default.
- W2891958946 isRetracted "false" @default.
- W2891958946 magId "2891958946" @default.
- W2891958946 workType "article" @default.