Matches in SemOpenAlex for { <https://semopenalex.org/work/W2892359699> ?p ?o ?g. }
- W2892359699 abstract "Teaching is critical to human society: it is with teaching that prospective students are educated and human civilization can be inherited and advanced. A good teacher not only provides his/her students with qualified teaching materials (e.g., textbooks), but also sets up appropriate objectives (e.g., course projects and exams) considering different situations of a student. When it comes to artificial intelligence, treating machine models as students, the loss that are optimized act as perfect counterparts of the objective set by the teacher. In this work, we explore the possibility of imitating human teaching behaviors by dynamically and automatically outputting appropriate loss to train machine models. Different from typical settings in which the loss function of a machine model is predefined and fixed, in our framework, the loss function of a machine model (we call it student) is defined by another machine model (we call it teacher). The ultimate goal of teacher model is cultivating the student to have better performance measured on development dataset. Towards that end, similar to human teaching, the teacher, a parametric model, dynamically outputs different loss that will be used and optimized by its student model at different training stages. We develop an efficient method for the teacher model that makes gradient based optimization possible, exempt of the ineffective solutions such as policy optimization. We name our method as learning to teach with dynamic loss functions (L2T-DLF for short). Extensive experiments on real world tasks including image classification and neural machine translation demonstrate that our method significantly improves the quality of various student models." @default.
- W2892359699 created "2018-09-27" @default.
- W2892359699 creator A5000889517 @default.
- W2892359699 creator A5007225481 @default.
- W2892359699 creator A5020025718 @default.
- W2892359699 creator A5021772140 @default.
- W2892359699 creator A5034685928 @default.
- W2892359699 creator A5070990160 @default.
- W2892359699 creator A5090379137 @default.
- W2892359699 date "2018-10-29" @default.
- W2892359699 modified "2023-09-22" @default.
- W2892359699 title "Learning to Teach with Dynamic Loss Functions" @default.
- W2892359699 cites W1487463423 @default.
- W2892359699 cites W1522301498 @default.
- W2892359699 cites W1525460779 @default.
- W2892359699 cites W1569296262 @default.
- W2892359699 cites W1868018859 @default.
- W2892359699 cites W1980287119 @default.
- W2892359699 cites W2016589492 @default.
- W2892359699 cites W2020764470 @default.
- W2892359699 cites W2101105183 @default.
- W2892359699 cites W2132083787 @default.
- W2892359699 cites W2132984949 @default.
- W2892359699 cites W2133564696 @default.
- W2892359699 cites W2134829400 @default.
- W2892359699 cites W2136208491 @default.
- W2892359699 cites W2148886952 @default.
- W2892359699 cites W2173248099 @default.
- W2892359699 cites W2176263492 @default.
- W2892359699 cites W2182055801 @default.
- W2892359699 cites W2194775991 @default.
- W2892359699 cites W2401231614 @default.
- W2892359699 cites W2409744450 @default.
- W2892359699 cites W2487501366 @default.
- W2892359699 cites W2553303224 @default.
- W2892359699 cites W2557026499 @default.
- W2892359699 cites W2566720494 @default.
- W2892359699 cites W2594830834 @default.
- W2892359699 cites W2613904329 @default.
- W2892359699 cites W2616877139 @default.
- W2892359699 cites W2626778328 @default.
- W2892359699 cites W2743473392 @default.
- W2892359699 cites W2766774033 @default.
- W2892359699 cites W2949888546 @default.
- W2892359699 cites W2950925480 @default.
- W2892359699 cites W2951785524 @default.
- W2892359699 cites W2952975409 @default.
- W2892359699 cites W2962784628 @default.
- W2892359699 cites W2963163972 @default.
- W2892359699 cites W2963446712 @default.
- W2892359699 cites W2963463964 @default.
- W2892359699 cites W2963789586 @default.
- W2892359699 cites W2964027370 @default.
- W2892359699 cites W3041866211 @default.
- W2892359699 hasPublicationYear "2018" @default.
- W2892359699 type Work @default.
- W2892359699 sameAs 2892359699 @default.
- W2892359699 citedByCount "3" @default.
- W2892359699 countsByYear W28923596992019 @default.
- W2892359699 countsByYear W28923596992021 @default.
- W2892359699 crossrefType "posted-content" @default.
- W2892359699 hasAuthorship W2892359699A5000889517 @default.
- W2892359699 hasAuthorship W2892359699A5007225481 @default.
- W2892359699 hasAuthorship W2892359699A5020025718 @default.
- W2892359699 hasAuthorship W2892359699A5021772140 @default.
- W2892359699 hasAuthorship W2892359699A5034685928 @default.
- W2892359699 hasAuthorship W2892359699A5070990160 @default.
- W2892359699 hasAuthorship W2892359699A5090379137 @default.
- W2892359699 hasConcept C105795698 @default.
- W2892359699 hasConcept C117251300 @default.
- W2892359699 hasConcept C119857082 @default.
- W2892359699 hasConcept C14036430 @default.
- W2892359699 hasConcept C145420912 @default.
- W2892359699 hasConcept C154945302 @default.
- W2892359699 hasConcept C15744967 @default.
- W2892359699 hasConcept C177264268 @default.
- W2892359699 hasConcept C199360897 @default.
- W2892359699 hasConcept C24574437 @default.
- W2892359699 hasConcept C33923547 @default.
- W2892359699 hasConcept C41008148 @default.
- W2892359699 hasConcept C78458016 @default.
- W2892359699 hasConcept C86803240 @default.
- W2892359699 hasConceptScore W2892359699C105795698 @default.
- W2892359699 hasConceptScore W2892359699C117251300 @default.
- W2892359699 hasConceptScore W2892359699C119857082 @default.
- W2892359699 hasConceptScore W2892359699C14036430 @default.
- W2892359699 hasConceptScore W2892359699C145420912 @default.
- W2892359699 hasConceptScore W2892359699C154945302 @default.
- W2892359699 hasConceptScore W2892359699C15744967 @default.
- W2892359699 hasConceptScore W2892359699C177264268 @default.
- W2892359699 hasConceptScore W2892359699C199360897 @default.
- W2892359699 hasConceptScore W2892359699C24574437 @default.
- W2892359699 hasConceptScore W2892359699C33923547 @default.
- W2892359699 hasConceptScore W2892359699C41008148 @default.
- W2892359699 hasConceptScore W2892359699C78458016 @default.
- W2892359699 hasConceptScore W2892359699C86803240 @default.
- W2892359699 hasLocation W28923596991 @default.
- W2892359699 hasOpenAccess W2892359699 @default.
- W2892359699 hasPrimaryLocation W28923596991 @default.
- W2892359699 hasRelatedWork W193132246 @default.