Matches in SemOpenAlex for { <https://semopenalex.org/work/W2899243567> ?p ?o ?g. }
- W2899243567 abstract "This paper addresses the problem of incremental domain adaptation (IDA) in natural language processing (NLP). We assume each domain comes one after another, and that we could only access data in the current domain. The goal of IDA is to build a unified model performing well on all the domains that we have encountered. We adopt the recurrent neural network (RNN) widely used in NLP, but augment it with a directly parameterized memory bank, which is retrieved by an attention mechanism at each step of RNN transition. The memory bank provides a natural way of IDA: when adapting our model to a new domain, we progressively add new slots to the memory bank, which increases the number of parameters, and thus the model capacity. We learn the new memory slots and fine-tune existing parameters by back-propagation. Experimental results show that our approach achieves significantly better performance than fine-tuning alone. Compared with expanding hidden states, our approach is more robust for old domains, shown by both empirical and theoretical results. Our model also outperforms previous work of IDA including elastic weight consolidation and progressive neural networks in the experiments." @default.
- W2899243567 created "2018-11-09" @default.
- W2899243567 creator A5018267778 @default.
- W2899243567 creator A5024821632 @default.
- W2899243567 creator A5050467922 @default.
- W2899243567 creator A5059577211 @default.
- W2899243567 creator A5081610684 @default.
- W2899243567 creator A5086603207 @default.
- W2899243567 date "2018-11-01" @default.
- W2899243567 modified "2023-09-27" @default.
- W2899243567 title "Progressive Memory Banks for Incremental Domain Adaptation" @default.
- W2899243567 cites W1731081199 @default.
- W2899243567 cites W1793121960 @default.
- W2899243567 cites W2015227583 @default.
- W2899243567 cites W2064675550 @default.
- W2899243567 cites W2115701093 @default.
- W2899243567 cites W2125320996 @default.
- W2899243567 cites W2130942839 @default.
- W2899243567 cites W2157331557 @default.
- W2899243567 cites W2250539671 @default.
- W2899243567 cites W2473930607 @default.
- W2899243567 cites W2560647685 @default.
- W2899243567 cites W2605043629 @default.
- W2899243567 cites W2737492962 @default.
- W2899243567 cites W2769216919 @default.
- W2899243567 cites W2804033308 @default.
- W2899243567 cites W2890969459 @default.
- W2899243567 cites W2892234010 @default.
- W2899243567 cites W2913699173 @default.
- W2899243567 cites W2915446515 @default.
- W2899243567 cites W2945383715 @default.
- W2899243567 cites W2948743095 @default.
- W2899243567 cites W2952631604 @default.
- W2899243567 cites W2952677972 @default.
- W2899243567 cites W2962724315 @default.
- W2899243567 cites W2962854379 @default.
- W2899243567 cites W2962897020 @default.
- W2899243567 cites W2963289713 @default.
- W2899243567 cites W2963448850 @default.
- W2899243567 cites W2963491014 @default.
- W2899243567 cites W2963540014 @default.
- W2899243567 cites W2963789888 @default.
- W2899243567 cites W2963790827 @default.
- W2899243567 cites W2963846996 @default.
- W2899243567 cites W2963850662 @default.
- W2899243567 cites W2963903950 @default.
- W2899243567 cites W2964059756 @default.
- W2899243567 cites W2964088867 @default.
- W2899243567 cites W2964189064 @default.
- W2899243567 cites W2964352358 @default.
- W2899243567 cites W1857789879 @default.
- W2899243567 cites W2426267443 @default.
- W2899243567 hasPublicationYear "2018" @default.
- W2899243567 type Work @default.
- W2899243567 sameAs 2899243567 @default.
- W2899243567 citedByCount "1" @default.
- W2899243567 countsByYear W28992435672020 @default.
- W2899243567 crossrefType "posted-content" @default.
- W2899243567 hasAuthorship W2899243567A5018267778 @default.
- W2899243567 hasAuthorship W2899243567A5024821632 @default.
- W2899243567 hasAuthorship W2899243567A5050467922 @default.
- W2899243567 hasAuthorship W2899243567A5059577211 @default.
- W2899243567 hasAuthorship W2899243567A5081610684 @default.
- W2899243567 hasAuthorship W2899243567A5086603207 @default.
- W2899243567 hasConcept C11413529 @default.
- W2899243567 hasConcept C119857082 @default.
- W2899243567 hasConcept C120665830 @default.
- W2899243567 hasConcept C121332964 @default.
- W2899243567 hasConcept C121955636 @default.
- W2899243567 hasConcept C134306372 @default.
- W2899243567 hasConcept C139807058 @default.
- W2899243567 hasConcept C144133560 @default.
- W2899243567 hasConcept C147168706 @default.
- W2899243567 hasConcept C154945302 @default.
- W2899243567 hasConcept C165464430 @default.
- W2899243567 hasConcept C2776014549 @default.
- W2899243567 hasConcept C2776434776 @default.
- W2899243567 hasConcept C33923547 @default.
- W2899243567 hasConcept C36503486 @default.
- W2899243567 hasConcept C41008148 @default.
- W2899243567 hasConcept C50644808 @default.
- W2899243567 hasConcept C95623464 @default.
- W2899243567 hasConceptScore W2899243567C11413529 @default.
- W2899243567 hasConceptScore W2899243567C119857082 @default.
- W2899243567 hasConceptScore W2899243567C120665830 @default.
- W2899243567 hasConceptScore W2899243567C121332964 @default.
- W2899243567 hasConceptScore W2899243567C121955636 @default.
- W2899243567 hasConceptScore W2899243567C134306372 @default.
- W2899243567 hasConceptScore W2899243567C139807058 @default.
- W2899243567 hasConceptScore W2899243567C144133560 @default.
- W2899243567 hasConceptScore W2899243567C147168706 @default.
- W2899243567 hasConceptScore W2899243567C154945302 @default.
- W2899243567 hasConceptScore W2899243567C165464430 @default.
- W2899243567 hasConceptScore W2899243567C2776014549 @default.
- W2899243567 hasConceptScore W2899243567C2776434776 @default.
- W2899243567 hasConceptScore W2899243567C33923547 @default.
- W2899243567 hasConceptScore W2899243567C36503486 @default.
- W2899243567 hasConceptScore W2899243567C41008148 @default.
- W2899243567 hasConceptScore W2899243567C50644808 @default.
- W2899243567 hasConceptScore W2899243567C95623464 @default.