Matches in SemOpenAlex for { <https://semopenalex.org/work/W4310282783> ?p ?o ?g. }
Showing items 1 to 85 of
85
with 100 items per page.
- W4310282783 abstract "We present XKD, a novel self-supervised framework to learn meaningful representations from unlabelled video clips. XKD is trained with two pseudo tasks. First, masked data reconstruction is performed to learn individual representations from audio and visual streams. Next, self-supervised cross-modal knowledge distillation is performed between the two modalities through teacher-student setups to learn complementary information. To identify the most effective information to transfer and also to tackle the domain gap between audio and visual modalities which could hinder knowledge transfer, we introduce a domain alignment and feature refinement strategy for effective cross-modal knowledge distillation. Lastly, to develop a general-purpose network capable of handling both audio and visual streams, modality-agnostic variants of our proposed framework are introduced, which use the same backbone for both audio and visual modalities. Our proposed cross-modal knowledge distillation improves linear evaluation top-1 accuracy of video action classification by 8.6% on UCF101, 8.2% on HMDB51, 13.9% on Kinetics-Sound, and 15.7% on Kinetics400. Additionally, our modality-agnostic variant shows promising results in developing a general-purpose network capable of learning both data streams for solving different downstream tasks." @default.
- W4310282783 created "2022-11-30" @default.
- W4310282783 creator A5039812985 @default.
- W4310282783 creator A5067632980 @default.
- W4310282783 date "2022-11-25" @default.
- W4310282783 modified "2023-09-25" @default.
- W4310282783 title "XKD: Cross-modal Knowledge Distillation with Domain Alignment for Video Representation Learning" @default.
- W4310282783 doi "https://doi.org/10.48550/arxiv.2211.13929" @default.
- W4310282783 hasPublicationYear "2022" @default.
- W4310282783 type Work @default.
- W4310282783 citedByCount "0" @default.
- W4310282783 crossrefType "posted-content" @default.
- W4310282783 hasAuthorship W4310282783A5039812985 @default.
- W4310282783 hasAuthorship W4310282783A5067632980 @default.
- W4310282783 hasBestOaLocation W43102827831 @default.
- W4310282783 hasConcept C119857082 @default.
- W4310282783 hasConcept C134306372 @default.
- W4310282783 hasConcept C138885662 @default.
- W4310282783 hasConcept C144024400 @default.
- W4310282783 hasConcept C153180895 @default.
- W4310282783 hasConcept C154945302 @default.
- W4310282783 hasConcept C17744445 @default.
- W4310282783 hasConcept C178790620 @default.
- W4310282783 hasConcept C185592680 @default.
- W4310282783 hasConcept C188027245 @default.
- W4310282783 hasConcept C199539241 @default.
- W4310282783 hasConcept C204030448 @default.
- W4310282783 hasConcept C204321447 @default.
- W4310282783 hasConcept C207685749 @default.
- W4310282783 hasConcept C2776359362 @default.
- W4310282783 hasConcept C2776401178 @default.
- W4310282783 hasConcept C2779903281 @default.
- W4310282783 hasConcept C2780226545 @default.
- W4310282783 hasConcept C3017588708 @default.
- W4310282783 hasConcept C33923547 @default.
- W4310282783 hasConcept C36289849 @default.
- W4310282783 hasConcept C36503486 @default.
- W4310282783 hasConcept C41008148 @default.
- W4310282783 hasConcept C41895202 @default.
- W4310282783 hasConcept C49774154 @default.
- W4310282783 hasConcept C71139939 @default.
- W4310282783 hasConcept C94625758 @default.
- W4310282783 hasConceptScore W4310282783C119857082 @default.
- W4310282783 hasConceptScore W4310282783C134306372 @default.
- W4310282783 hasConceptScore W4310282783C138885662 @default.
- W4310282783 hasConceptScore W4310282783C144024400 @default.
- W4310282783 hasConceptScore W4310282783C153180895 @default.
- W4310282783 hasConceptScore W4310282783C154945302 @default.
- W4310282783 hasConceptScore W4310282783C17744445 @default.
- W4310282783 hasConceptScore W4310282783C178790620 @default.
- W4310282783 hasConceptScore W4310282783C185592680 @default.
- W4310282783 hasConceptScore W4310282783C188027245 @default.
- W4310282783 hasConceptScore W4310282783C199539241 @default.
- W4310282783 hasConceptScore W4310282783C204030448 @default.
- W4310282783 hasConceptScore W4310282783C204321447 @default.
- W4310282783 hasConceptScore W4310282783C207685749 @default.
- W4310282783 hasConceptScore W4310282783C2776359362 @default.
- W4310282783 hasConceptScore W4310282783C2776401178 @default.
- W4310282783 hasConceptScore W4310282783C2779903281 @default.
- W4310282783 hasConceptScore W4310282783C2780226545 @default.
- W4310282783 hasConceptScore W4310282783C3017588708 @default.
- W4310282783 hasConceptScore W4310282783C33923547 @default.
- W4310282783 hasConceptScore W4310282783C36289849 @default.
- W4310282783 hasConceptScore W4310282783C36503486 @default.
- W4310282783 hasConceptScore W4310282783C41008148 @default.
- W4310282783 hasConceptScore W4310282783C41895202 @default.
- W4310282783 hasConceptScore W4310282783C49774154 @default.
- W4310282783 hasConceptScore W4310282783C71139939 @default.
- W4310282783 hasConceptScore W4310282783C94625758 @default.
- W4310282783 hasLocation W43102827831 @default.
- W4310282783 hasOpenAccess W4310282783 @default.
- W4310282783 hasPrimaryLocation W43102827831 @default.
- W4310282783 hasRelatedWork W2546190447 @default.
- W4310282783 hasRelatedWork W2949074159 @default.
- W4310282783 hasRelatedWork W2952745240 @default.
- W4310282783 hasRelatedWork W3107474891 @default.
- W4310282783 hasRelatedWork W3142456083 @default.
- W4310282783 hasRelatedWork W3211385060 @default.
- W4310282783 hasRelatedWork W4225369406 @default.
- W4310282783 hasRelatedWork W4298715519 @default.
- W4310282783 hasRelatedWork W4301143707 @default.
- W4310282783 hasRelatedWork W4306353150 @default.
- W4310282783 isParatext "false" @default.
- W4310282783 isRetracted "false" @default.
- W4310282783 workType "article" @default.