Matches in SemOpenAlex for { <https://semopenalex.org/work/W4375868968> ?p ?o ?g. }
- W4375868968 abstract "Neural network-based speaker recognition has achieved significant improvement in recent years. A robust speaker representation learns meaningful knowledge from both hard and easy samples in the training set to achieve good performance. However, noisy samples (i.e., with wrong labels) in the training set induce confusion and cause the network to learn the incorrect representation. In this paper, we propose a two-step audio-visual deep cleansing framework to eliminate the effect of noisy labels in speaker representation learning. This framework contains a coarse-grained cleansing step to search for the complex samples, followed by a fine-grained cleansing step to filter out the noisy labels. Our study starts from an efficient audio-visual speaker recognition system, which achieves a close to perfect equal-error-rate (EER) of 0.01%, 0.07% and 0.13% on the Vox-O, E and H test sets. With the proposed multi-modal cleansing mechanism, four different speaker recognition networks achieve an average improvement of 5.9%. Code has been made available at: https://github.com/TaoRuijie/AVCleanse." @default.
- W4375868968 created "2023-05-10" @default.
- W4375868968 creator A5000218886 @default.
- W4375868968 creator A5004287909 @default.
- W4375868968 creator A5026034735 @default.
- W4375868968 creator A5032690182 @default.
- W4375868968 date "2023-06-04" @default.
- W4375868968 modified "2023-09-27" @default.
- W4375868968 title "Speaker Recognition with Two-Step Multi-Modal Deep Cleansing" @default.
- W4375868968 cites W2696967604 @default.
- W4375868968 cites W2726515241 @default.
- W4375868968 cites W2808631503 @default.
- W4375868968 cites W2890964092 @default.
- W4375868968 cites W2902299888 @default.
- W4375868968 cites W2921475306 @default.
- W4375868968 cites W2962819150 @default.
- W4375868968 cites W2963496125 @default.
- W4375868968 cites W2969985801 @default.
- W4375868968 cites W2988966271 @default.
- W4375868968 cites W3013020904 @default.
- W4375868968 cites W3015598461 @default.
- W4375868968 cites W3024869864 @default.
- W4375868968 cites W3099206234 @default.
- W4375868968 cites W3101998545 @default.
- W4375868968 cites W3141035251 @default.
- W4375868968 cites W3174649707 @default.
- W4375868968 cites W3205635414 @default.
- W4375868968 cites W3206287456 @default.
- W4375868968 cites W3217435001 @default.
- W4375868968 cites W4221154746 @default.
- W4375868968 cites W4239510810 @default.
- W4375868968 cites W4312095872 @default.
- W4375868968 cites W3096936581 @default.
- W4375868968 doi "https://doi.org/10.1109/icassp49357.2023.10096814" @default.
- W4375868968 hasPublicationYear "2023" @default.
- W4375868968 type Work @default.
- W4375868968 citedByCount "0" @default.
- W4375868968 crossrefType "proceedings-article" @default.
- W4375868968 hasAuthorship W4375868968A5000218886 @default.
- W4375868968 hasAuthorship W4375868968A5004287909 @default.
- W4375868968 hasAuthorship W4375868968A5026034735 @default.
- W4375868968 hasAuthorship W4375868968A5032690182 @default.
- W4375868968 hasBestOaLocation W43758689681 @default.
- W4375868968 hasConcept C106131492 @default.
- W4375868968 hasConcept C133892786 @default.
- W4375868968 hasConcept C153180895 @default.
- W4375868968 hasConcept C154945302 @default.
- W4375868968 hasConcept C162324750 @default.
- W4375868968 hasConcept C176217482 @default.
- W4375868968 hasConcept C177264268 @default.
- W4375868968 hasConcept C17744445 @default.
- W4375868968 hasConcept C185592680 @default.
- W4375868968 hasConcept C188027245 @default.
- W4375868968 hasConcept C199360897 @default.
- W4375868968 hasConcept C199539241 @default.
- W4375868968 hasConcept C21547014 @default.
- W4375868968 hasConcept C24756922 @default.
- W4375868968 hasConcept C2776359362 @default.
- W4375868968 hasConcept C2776760102 @default.
- W4375868968 hasConcept C28490314 @default.
- W4375868968 hasConcept C31972630 @default.
- W4375868968 hasConcept C40969351 @default.
- W4375868968 hasConcept C41008148 @default.
- W4375868968 hasConcept C42199009 @default.
- W4375868968 hasConcept C50644808 @default.
- W4375868968 hasConcept C71139939 @default.
- W4375868968 hasConcept C94625758 @default.
- W4375868968 hasConceptScore W4375868968C106131492 @default.
- W4375868968 hasConceptScore W4375868968C133892786 @default.
- W4375868968 hasConceptScore W4375868968C153180895 @default.
- W4375868968 hasConceptScore W4375868968C154945302 @default.
- W4375868968 hasConceptScore W4375868968C162324750 @default.
- W4375868968 hasConceptScore W4375868968C176217482 @default.
- W4375868968 hasConceptScore W4375868968C177264268 @default.
- W4375868968 hasConceptScore W4375868968C17744445 @default.
- W4375868968 hasConceptScore W4375868968C185592680 @default.
- W4375868968 hasConceptScore W4375868968C188027245 @default.
- W4375868968 hasConceptScore W4375868968C199360897 @default.
- W4375868968 hasConceptScore W4375868968C199539241 @default.
- W4375868968 hasConceptScore W4375868968C21547014 @default.
- W4375868968 hasConceptScore W4375868968C24756922 @default.
- W4375868968 hasConceptScore W4375868968C2776359362 @default.
- W4375868968 hasConceptScore W4375868968C2776760102 @default.
- W4375868968 hasConceptScore W4375868968C28490314 @default.
- W4375868968 hasConceptScore W4375868968C31972630 @default.
- W4375868968 hasConceptScore W4375868968C40969351 @default.
- W4375868968 hasConceptScore W4375868968C41008148 @default.
- W4375868968 hasConceptScore W4375868968C42199009 @default.
- W4375868968 hasConceptScore W4375868968C50644808 @default.
- W4375868968 hasConceptScore W4375868968C71139939 @default.
- W4375868968 hasConceptScore W4375868968C94625758 @default.
- W4375868968 hasFunder F4320321001 @default.
- W4375868968 hasFunder F4320322170 @default.
- W4375868968 hasFunder F4320331102 @default.
- W4375868968 hasLocation W43758689681 @default.
- W4375868968 hasLocation W43758689682 @default.
- W4375868968 hasOpenAccess W4375868968 @default.
- W4375868968 hasPrimaryLocation W43758689681 @default.
- W4375868968 hasRelatedWork W1960256358 @default.
- W4375868968 hasRelatedWork W2112059504 @default.