Matches in SemOpenAlex for { <https://semopenalex.org/work/W4225766285> ?p ?o ?g. }
- W4225766285 endingPage "738" @default.
- W4225766285 startingPage "717" @default.
- W4225766285 abstract "Recent studies on learning with noisy labels have shown remarkable performance by exploiting a small clean dataset. In particular, model agnostic meta-learning-based label correction methods further improve performance by correcting noisy labels on the fly. However, there is no safeguard on the label miscorrection, resulting in unavoidable performance degradation. Moreover, every training step requires at least three back-propagations, significantly slowing down the training speed. To mitigate these issues, we propose a robust and efficient method, FasTEN, which learns a label transition matrix on the fly. Employing the transition matrix makes the classifier skeptical about all the corrected samples, which alleviates the miscorrection issue. We also introduce a two-head architecture to efficiently estimate the label transition matrix every iteration within a single back-propagation, so that the estimated matrix closely follows the shifting noise distribution induced by label correction. Extensive experiments demonstrate that our FasTEN shows the best performance in training efficiency while having comparable or better accuracy than existing methods, especially achieving state-of-the-art performance in a real-world noisy dataset, Clothing1M." @default.
- W4225766285 created "2022-05-05" @default.
- W4225766285 creator A5023799668 @default.
- W4225766285 creator A5043877199 @default.
- W4225766285 creator A5068698319 @default.
- W4225766285 creator A5075636997 @default.
- W4225766285 date "2022-01-01" @default.
- W4225766285 modified "2023-10-17" @default.
- W4225766285 title "Learning with Noisy Labels by Efficient Transition Matrix Estimation to Combat Label Miscorrection" @default.
- W4225766285 cites W1514928307 @default.
- W4225766285 cites W1969623397 @default.
- W4225766285 cites W1982032418 @default.
- W4225766285 cites W2021646111 @default.
- W4225766285 cites W2102605133 @default.
- W4225766285 cites W2108598243 @default.
- W4225766285 cites W2145287260 @default.
- W4225766285 cites W2148143831 @default.
- W4225766285 cites W2149298154 @default.
- W4225766285 cites W2194775991 @default.
- W4225766285 cites W2403681572 @default.
- W4225766285 cites W2554864439 @default.
- W4225766285 cites W2577784528 @default.
- W4225766285 cites W2905290040 @default.
- W4225766285 cites W2962762068 @default.
- W4225766285 cites W2963351448 @default.
- W4225766285 cites W2963516811 @default.
- W4225766285 cites W2963697299 @default.
- W4225766285 cites W2963772355 @default.
- W4225766285 cites W2963909971 @default.
- W4225766285 cites W2964155802 @default.
- W4225766285 cites W2964273174 @default.
- W4225766285 cites W2964274690 @default.
- W4225766285 cites W2964292098 @default.
- W4225766285 cites W2967052791 @default.
- W4225766285 cites W2978625989 @default.
- W4225766285 cites W2979302305 @default.
- W4225766285 cites W2981873476 @default.
- W4225766285 cites W2990019157 @default.
- W4225766285 cites W2997199946 @default.
- W4225766285 cites W2998046109 @default.
- W4225766285 cites W3035716296 @default.
- W4225766285 cites W3046688540 @default.
- W4225766285 cites W3095319910 @default.
- W4225766285 cites W3110687497 @default.
- W4225766285 cites W3119583506 @default.
- W4225766285 cites W3170414933 @default.
- W4225766285 cites W3171635822 @default.
- W4225766285 cites W3173623546 @default.
- W4225766285 cites W3173733742 @default.
- W4225766285 cites W3175103763 @default.
- W4225766285 cites W3177200443 @default.
- W4225766285 cites W4230674625 @default.
- W4225766285 cites W4245577611 @default.
- W4225766285 cites W4293409613 @default.
- W4225766285 doi "https://doi.org/10.1007/978-3-031-19806-9_41" @default.
- W4225766285 hasPublicationYear "2022" @default.
- W4225766285 type Work @default.
- W4225766285 citedByCount "1" @default.
- W4225766285 countsByYear W42257662852023 @default.
- W4225766285 crossrefType "book-chapter" @default.
- W4225766285 hasAuthorship W4225766285A5023799668 @default.
- W4225766285 hasAuthorship W4225766285A5043877199 @default.
- W4225766285 hasAuthorship W4225766285A5068698319 @default.
- W4225766285 hasAuthorship W4225766285A5075636997 @default.
- W4225766285 hasBestOaLocation W42257662852 @default.
- W4225766285 hasConcept C104317684 @default.
- W4225766285 hasConcept C106487976 @default.
- W4225766285 hasConcept C111919701 @default.
- W4225766285 hasConcept C11413529 @default.
- W4225766285 hasConcept C115961682 @default.
- W4225766285 hasConcept C119857082 @default.
- W4225766285 hasConcept C153180895 @default.
- W4225766285 hasConcept C154945302 @default.
- W4225766285 hasConcept C159985019 @default.
- W4225766285 hasConcept C185592680 @default.
- W4225766285 hasConcept C192562407 @default.
- W4225766285 hasConcept C2781020372 @default.
- W4225766285 hasConcept C41008148 @default.
- W4225766285 hasConcept C49555168 @default.
- W4225766285 hasConcept C55493867 @default.
- W4225766285 hasConcept C63479239 @default.
- W4225766285 hasConcept C95623464 @default.
- W4225766285 hasConcept C98763669 @default.
- W4225766285 hasConcept C99498987 @default.
- W4225766285 hasConceptScore W4225766285C104317684 @default.
- W4225766285 hasConceptScore W4225766285C106487976 @default.
- W4225766285 hasConceptScore W4225766285C111919701 @default.
- W4225766285 hasConceptScore W4225766285C11413529 @default.
- W4225766285 hasConceptScore W4225766285C115961682 @default.
- W4225766285 hasConceptScore W4225766285C119857082 @default.
- W4225766285 hasConceptScore W4225766285C153180895 @default.
- W4225766285 hasConceptScore W4225766285C154945302 @default.
- W4225766285 hasConceptScore W4225766285C159985019 @default.
- W4225766285 hasConceptScore W4225766285C185592680 @default.
- W4225766285 hasConceptScore W4225766285C192562407 @default.
- W4225766285 hasConceptScore W4225766285C2781020372 @default.
- W4225766285 hasConceptScore W4225766285C41008148 @default.
- W4225766285 hasConceptScore W4225766285C49555168 @default.