Matches in SemOpenAlex for { <https://semopenalex.org/work/W3137477868> ?p ?o ?g. }
- W3137477868 abstract "Nowadays, we have witnessed the early progress on learning the association between voice and face automatically, which brings a new wave of studies to the computer vision community. However, most of the prior arts along this line (a) merely adopt local information to perform modality alignment and (b) ignore the diversity of learning difficulty across different subjects. In this paper, we propose a novel framework to jointly address the above-mentioned issues. Targeting at (a), we propose a two-level modality alignment loss where both global and local information are considered. Compared with the existing methods, we introduce a global loss into the modality alignment process. The global component of the loss is driven by the identity classification. Theoretically, we show that minimizing the loss could maximize the distance between embeddings across different identities while minimizing the distance between embeddings belonging to the same identity, in a global sense (instead of a mini-batch). Targeting at (b), we propose a dynamic reweighting scheme to better explore the hard but valuable identities while filtering out the unlearnable identities. Experiments show that the proposed method outperforms the previous methods in multiple settings, including voice-face matching, verification and retrieval." @default.
- W3137477868 created "2021-03-29" @default.
- W3137477868 creator A5009753955 @default.
- W3137477868 creator A5018138384 @default.
- W3137477868 creator A5028597017 @default.
- W3137477868 creator A5053213796 @default.
- W3137477868 creator A5059053033 @default.
- W3137477868 creator A5089092566 @default.
- W3137477868 date "2021-03-12" @default.
- W3137477868 modified "2023-09-24" @default.
- W3137477868 title "Seeking the Shape of Sound: An Adaptive Framework for Learning Voice-Face Association" @default.
- W3137477868 cites W1995945562 @default.
- W3137477868 cites W2066078528 @default.
- W3137477868 cites W2077881910 @default.
- W3137477868 cites W2106053110 @default.
- W3137477868 cites W2130556178 @default.
- W3137477868 cites W2138621090 @default.
- W3137477868 cites W2179042386 @default.
- W3137477868 cites W2227324914 @default.
- W3137477868 cites W2325939864 @default.
- W3137477868 cites W2515770085 @default.
- W3137477868 cites W2594061220 @default.
- W3137477868 cites W2597603852 @default.
- W3137477868 cites W2808631503 @default.
- W3137477868 cites W2897663715 @default.
- W3137477868 cites W2951237705 @default.
- W3137477868 cites W2962918445 @default.
- W3137477868 cites W2963070594 @default.
- W3137477868 cites W2963113370 @default.
- W3137477868 cites W2963351448 @default.
- W3137477868 cites W2963371670 @default.
- W3137477868 cites W2963420686 @default.
- W3137477868 cites W2963507712 @default.
- W3137477868 cites W2963516811 @default.
- W3137477868 cites W2963801643 @default.
- W3137477868 cites W2963887950 @default.
- W3137477868 cites W2970903655 @default.
- W3137477868 cites W2979157532 @default.
- W3137477868 cites W2991096973 @default.
- W3137477868 cites W2995521531 @default.
- W3137477868 cites W3034303554 @default.
- W3137477868 cites W3034725201 @default.
- W3137477868 cites W3037563025 @default.
- W3137477868 cites W3128564814 @default.
- W3137477868 doi "https://doi.org/10.48550/arxiv.2103.07293" @default.
- W3137477868 hasPublicationYear "2021" @default.
- W3137477868 type Work @default.
- W3137477868 sameAs 3137477868 @default.
- W3137477868 citedByCount "0" @default.
- W3137477868 crossrefType "posted-content" @default.
- W3137477868 hasAuthorship W3137477868A5009753955 @default.
- W3137477868 hasAuthorship W3137477868A5018138384 @default.
- W3137477868 hasAuthorship W3137477868A5028597017 @default.
- W3137477868 hasAuthorship W3137477868A5053213796 @default.
- W3137477868 hasAuthorship W3137477868A5059053033 @default.
- W3137477868 hasAuthorship W3137477868A5089092566 @default.
- W3137477868 hasBestOaLocation W31374778681 @default.
- W3137477868 hasConcept C105795698 @default.
- W3137477868 hasConcept C111919701 @default.
- W3137477868 hasConcept C119857082 @default.
- W3137477868 hasConcept C121332964 @default.
- W3137477868 hasConcept C134306372 @default.
- W3137477868 hasConcept C138885662 @default.
- W3137477868 hasConcept C142853389 @default.
- W3137477868 hasConcept C154945302 @default.
- W3137477868 hasConcept C15744967 @default.
- W3137477868 hasConcept C165064840 @default.
- W3137477868 hasConcept C24890656 @default.
- W3137477868 hasConcept C2778355321 @default.
- W3137477868 hasConcept C2779304628 @default.
- W3137477868 hasConcept C2780226545 @default.
- W3137477868 hasConcept C33923547 @default.
- W3137477868 hasConcept C41008148 @default.
- W3137477868 hasConcept C41895202 @default.
- W3137477868 hasConcept C542102704 @default.
- W3137477868 hasConcept C77618280 @default.
- W3137477868 hasConcept C98045186 @default.
- W3137477868 hasConceptScore W3137477868C105795698 @default.
- W3137477868 hasConceptScore W3137477868C111919701 @default.
- W3137477868 hasConceptScore W3137477868C119857082 @default.
- W3137477868 hasConceptScore W3137477868C121332964 @default.
- W3137477868 hasConceptScore W3137477868C134306372 @default.
- W3137477868 hasConceptScore W3137477868C138885662 @default.
- W3137477868 hasConceptScore W3137477868C142853389 @default.
- W3137477868 hasConceptScore W3137477868C154945302 @default.
- W3137477868 hasConceptScore W3137477868C15744967 @default.
- W3137477868 hasConceptScore W3137477868C165064840 @default.
- W3137477868 hasConceptScore W3137477868C24890656 @default.
- W3137477868 hasConceptScore W3137477868C2778355321 @default.
- W3137477868 hasConceptScore W3137477868C2779304628 @default.
- W3137477868 hasConceptScore W3137477868C2780226545 @default.
- W3137477868 hasConceptScore W3137477868C33923547 @default.
- W3137477868 hasConceptScore W3137477868C41008148 @default.
- W3137477868 hasConceptScore W3137477868C41895202 @default.
- W3137477868 hasConceptScore W3137477868C542102704 @default.
- W3137477868 hasConceptScore W3137477868C77618280 @default.
- W3137477868 hasConceptScore W3137477868C98045186 @default.
- W3137477868 hasLocation W31374778681 @default.
- W3137477868 hasOpenAccess W3137477868 @default.
- W3137477868 hasPrimaryLocation W31374778681 @default.