Matches in SemOpenAlex for { <https://semopenalex.org/work/W4385570810> ?p ?o ?g. }
Showing items 1 to 85 of
85
with 100 items per page.
- W4385570810 abstract "Bias in machine learning models can be an issue when the models are trained on particular types of data that do not generalize well, causing under performance in certain groups of users. In this work, we focus on reducing the bias related to new customers in a digital voice assistant system. It is observed that natural language understanding models often have lower performance when dealing with requests coming from new users rather than experienced users. To mitigate this problem, we propose a framework that consists of two phases (1) a fixing phase with four active learning strategies used to identify important samples coming from new users, and (2) a self training phase where a teacher model trained from the first phase is used to annotate semi-supervised samples to expand the training data with relevant cohort utterances. We explain practical strategies that involve an identification of representative cohort-based samples through density clustering as well as employing implicit customer feedbacks to improve new customers’ experience. We demonstrate the effectiveness of our approach in a real world large scale voice assistant system for two languages, German and French through both offline experiments as well as A/B testings." @default.
- W4385570810 created "2023-08-05" @default.
- W4385570810 creator A5050249771 @default.
- W4385570810 creator A5078854539 @default.
- W4385570810 creator A5085695654 @default.
- W4385570810 creator A5089660384 @default.
- W4385570810 date "2023-01-01" @default.
- W4385570810 modified "2023-10-14" @default.
- W4385570810 title "Reducing cohort bias in natural language understanding systems with targeted self-training scheme" @default.
- W4385570810 doi "https://doi.org/10.18653/v1/2023.acl-industry.53" @default.
- W4385570810 hasPublicationYear "2023" @default.
- W4385570810 type Work @default.
- W4385570810 citedByCount "0" @default.
- W4385570810 crossrefType "proceedings-article" @default.
- W4385570810 hasAuthorship W4385570810A5050249771 @default.
- W4385570810 hasAuthorship W4385570810A5078854539 @default.
- W4385570810 hasAuthorship W4385570810A5085695654 @default.
- W4385570810 hasAuthorship W4385570810A5089660384 @default.
- W4385570810 hasBestOaLocation W43855708101 @default.
- W4385570810 hasConcept C105795698 @default.
- W4385570810 hasConcept C116834253 @default.
- W4385570810 hasConcept C119857082 @default.
- W4385570810 hasConcept C120665830 @default.
- W4385570810 hasConcept C121332964 @default.
- W4385570810 hasConcept C134306372 @default.
- W4385570810 hasConcept C136389625 @default.
- W4385570810 hasConcept C153294291 @default.
- W4385570810 hasConcept C154775046 @default.
- W4385570810 hasConcept C154945302 @default.
- W4385570810 hasConcept C166957645 @default.
- W4385570810 hasConcept C192209626 @default.
- W4385570810 hasConcept C195324797 @default.
- W4385570810 hasConcept C204321447 @default.
- W4385570810 hasConcept C2777211547 @default.
- W4385570810 hasConcept C2779439875 @default.
- W4385570810 hasConcept C33923547 @default.
- W4385570810 hasConcept C41008148 @default.
- W4385570810 hasConcept C50644808 @default.
- W4385570810 hasConcept C59822182 @default.
- W4385570810 hasConcept C72563966 @default.
- W4385570810 hasConcept C73555534 @default.
- W4385570810 hasConcept C77618280 @default.
- W4385570810 hasConcept C86803240 @default.
- W4385570810 hasConcept C95457728 @default.
- W4385570810 hasConceptScore W4385570810C105795698 @default.
- W4385570810 hasConceptScore W4385570810C116834253 @default.
- W4385570810 hasConceptScore W4385570810C119857082 @default.
- W4385570810 hasConceptScore W4385570810C120665830 @default.
- W4385570810 hasConceptScore W4385570810C121332964 @default.
- W4385570810 hasConceptScore W4385570810C134306372 @default.
- W4385570810 hasConceptScore W4385570810C136389625 @default.
- W4385570810 hasConceptScore W4385570810C153294291 @default.
- W4385570810 hasConceptScore W4385570810C154775046 @default.
- W4385570810 hasConceptScore W4385570810C154945302 @default.
- W4385570810 hasConceptScore W4385570810C166957645 @default.
- W4385570810 hasConceptScore W4385570810C192209626 @default.
- W4385570810 hasConceptScore W4385570810C195324797 @default.
- W4385570810 hasConceptScore W4385570810C204321447 @default.
- W4385570810 hasConceptScore W4385570810C2777211547 @default.
- W4385570810 hasConceptScore W4385570810C2779439875 @default.
- W4385570810 hasConceptScore W4385570810C33923547 @default.
- W4385570810 hasConceptScore W4385570810C41008148 @default.
- W4385570810 hasConceptScore W4385570810C50644808 @default.
- W4385570810 hasConceptScore W4385570810C59822182 @default.
- W4385570810 hasConceptScore W4385570810C72563966 @default.
- W4385570810 hasConceptScore W4385570810C73555534 @default.
- W4385570810 hasConceptScore W4385570810C77618280 @default.
- W4385570810 hasConceptScore W4385570810C86803240 @default.
- W4385570810 hasConceptScore W4385570810C95457728 @default.
- W4385570810 hasLocation W43855708101 @default.
- W4385570810 hasOpenAccess W4385570810 @default.
- W4385570810 hasPrimaryLocation W43855708101 @default.
- W4385570810 hasRelatedWork W159132833 @default.
- W4385570810 hasRelatedWork W2293457016 @default.
- W4385570810 hasRelatedWork W2602143361 @default.
- W4385570810 hasRelatedWork W2780329366 @default.
- W4385570810 hasRelatedWork W2977842567 @default.
- W4385570810 hasRelatedWork W3162567751 @default.
- W4385570810 hasRelatedWork W4221088574 @default.
- W4385570810 hasRelatedWork W4285260836 @default.
- W4385570810 hasRelatedWork W87581401 @default.
- W4385570810 hasRelatedWork W1872130062 @default.
- W4385570810 isParatext "false" @default.
- W4385570810 isRetracted "false" @default.
- W4385570810 workType "article" @default.