Matches in SemOpenAlex for { <https://semopenalex.org/work/W3174046974> ?p ?o ?g. }
- W3174046974 abstract "Multimodal representations and continual learning are two areas closely related to human intelligence. The former considers the learning of shared representation spaces where information from different modalities can be compared and integrated (we focus on cross-modal retrieval between language and visual representations). The latter studies how to prevent forgetting a previously learned task when learning a new one. While humans excel in these two aspects, deep neural networks are still quite limited. In this paper, we propose a combination of both problems into a continual cross-modal retrieval setting, where we study how the catastrophic interference caused by new tasks impacts the embedding spaces and their cross-modal alignment required for effective retrieval. We propose a general framework that decouples the training, indexing and querying stages. We also identify and study different factors that may lead to forgetting, and propose tools to alleviate it. We found that the indexing stage pays an important role and that simply avoiding reindexing the database with updated embedding networks can lead to significant gains. We evaluated our methods in two image-text retrieval datasets, obtaining significant gains with respect to the fine tuning baseline." @default.
- W3174046974 created "2021-07-05" @default.
- W3174046974 creator A5073531557 @default.
- W3174046974 creator A5079201773 @default.
- W3174046974 creator A5081872084 @default.
- W3174046974 date "2021-06-01" @default.
- W3174046974 modified "2023-10-01" @default.
- W3174046974 title "Continual learning in cross-modal retrieval" @default.
- W3174046974 cites W1682403713 @default.
- W3174046974 cites W1869500417 @default.
- W3174046974 cites W1964073652 @default.
- W3174046974 cites W1970055505 @default.
- W3174046974 cites W2025341678 @default.
- W3174046974 cites W2061524864 @default.
- W3174046974 cites W2116522068 @default.
- W3174046974 cites W2127589108 @default.
- W3174046974 cites W2138621090 @default.
- W3174046974 cites W2149557440 @default.
- W3174046974 cites W2266728343 @default.
- W3174046974 cites W2277195237 @default.
- W3174046974 cites W2388114291 @default.
- W3174046974 cites W2473930607 @default.
- W3174046974 cites W2560647685 @default.
- W3174046974 cites W2562417371 @default.
- W3174046974 cites W2605102252 @default.
- W3174046974 cites W2606473278 @default.
- W3174046974 cites W2619383789 @default.
- W3174046974 cites W2737041163 @default.
- W3174046974 cites W2963047368 @default.
- W3174046974 cites W2963586458 @default.
- W3174046974 cites W2963775347 @default.
- W3174046974 cites W2963788399 @default.
- W3174046974 cites W2964024144 @default.
- W3174046974 cites W2964189064 @default.
- W3174046974 cites W3035212740 @default.
- W3174046974 cites W3035284526 @default.
- W3174046974 cites W3035454331 @default.
- W3174046974 cites W3035497460 @default.
- W3174046974 cites W3035501943 @default.
- W3174046974 cites W3035644209 @default.
- W3174046974 cites W3046253174 @default.
- W3174046974 cites W3099206234 @default.
- W3174046974 cites W3103800629 @default.
- W3174046974 cites W3168433561 @default.
- W3174046974 cites W4210880854 @default.
- W3174046974 cites W877909479 @default.
- W3174046974 cites W92662927 @default.
- W3174046974 cites W2964181976 @default.
- W3174046974 doi "https://doi.org/10.1109/cvprw53098.2021.00402" @default.
- W3174046974 hasPublicationYear "2021" @default.
- W3174046974 type Work @default.
- W3174046974 sameAs 3174046974 @default.
- W3174046974 citedByCount "5" @default.
- W3174046974 countsByYear W31740469742022 @default.
- W3174046974 countsByYear W31740469742023 @default.
- W3174046974 crossrefType "proceedings-article" @default.
- W3174046974 hasAuthorship W3174046974A5073531557 @default.
- W3174046974 hasAuthorship W3174046974A5079201773 @default.
- W3174046974 hasAuthorship W3174046974A5081872084 @default.
- W3174046974 hasBestOaLocation W31740469742 @default.
- W3174046974 hasConcept C108583219 @default.
- W3174046974 hasConcept C111368507 @default.
- W3174046974 hasConcept C119857082 @default.
- W3174046974 hasConcept C120665830 @default.
- W3174046974 hasConcept C121332964 @default.
- W3174046974 hasConcept C12725497 @default.
- W3174046974 hasConcept C127313418 @default.
- W3174046974 hasConcept C138885662 @default.
- W3174046974 hasConcept C144024400 @default.
- W3174046974 hasConcept C154945302 @default.
- W3174046974 hasConcept C162324750 @default.
- W3174046974 hasConcept C17744445 @default.
- W3174046974 hasConcept C185592680 @default.
- W3174046974 hasConcept C187736073 @default.
- W3174046974 hasConcept C188027245 @default.
- W3174046974 hasConcept C192209626 @default.
- W3174046974 hasConcept C199539241 @default.
- W3174046974 hasConcept C23123220 @default.
- W3174046974 hasConcept C2776359362 @default.
- W3174046974 hasConcept C2776401178 @default.
- W3174046974 hasConcept C2779903281 @default.
- W3174046974 hasConcept C2780451532 @default.
- W3174046974 hasConcept C36289849 @default.
- W3174046974 hasConcept C41008148 @default.
- W3174046974 hasConcept C41608201 @default.
- W3174046974 hasConcept C41895202 @default.
- W3174046974 hasConcept C59404180 @default.
- W3174046974 hasConcept C71139939 @default.
- W3174046974 hasConcept C7149132 @default.
- W3174046974 hasConcept C75165309 @default.
- W3174046974 hasConcept C94625758 @default.
- W3174046974 hasConceptScore W3174046974C108583219 @default.
- W3174046974 hasConceptScore W3174046974C111368507 @default.
- W3174046974 hasConceptScore W3174046974C119857082 @default.
- W3174046974 hasConceptScore W3174046974C120665830 @default.
- W3174046974 hasConceptScore W3174046974C121332964 @default.
- W3174046974 hasConceptScore W3174046974C12725497 @default.
- W3174046974 hasConceptScore W3174046974C127313418 @default.
- W3174046974 hasConceptScore W3174046974C138885662 @default.
- W3174046974 hasConceptScore W3174046974C144024400 @default.