Matches in SemOpenAlex for { <https://semopenalex.org/work/W3216446504> ?p ?o ?g. }
- W3216446504 abstract "Mixup is a well-known data-dependent augmentation technique for DNNs, consisting of two sub-tasks: mixup generation and classification. However, the recent dominant online training method confines mixup to supervised learning (SL), and the objective of the generation sub-task is limited to selected sample pairs instead of the whole data manifold, which might cause trivial solutions. To overcome such limitations, we comprehensively study the objective of mixup generation and propose textbf{S}cenario-textbf{A}gnostic textbf{Mix}up (SAMix) for both SL and Self-supervised Learning (SSL) scenarios. Specifically, we hypothesize and verify the objective function of mixup generation as optimizing local smoothness between two mixed classes subject to global discrimination from other classes. Accordingly, we propose $eta$-balanced mixup loss for complementary learning of the two sub-objectives. Meanwhile, a label-free generation sub-network is designed, which effectively provides non-trivial mixup samples and improves transferable abilities. Moreover, to reduce the computational cost of online training, we further introduce a pre-trained version, SAMix$^mathcal{P}$, achieving more favorable efficiency and generalizability. Extensive experiments on nine SL and SSL benchmarks demonstrate the consistent superiority and versatility of SAMix compared with existing methods." @default.
- W3216446504 created "2021-12-06" @default.
- W3216446504 creator A5011760791 @default.
- W3216446504 creator A5033050983 @default.
- W3216446504 creator A5048295582 @default.
- W3216446504 creator A5055640195 @default.
- W3216446504 creator A5082786719 @default.
- W3216446504 date "2021-11-30" @default.
- W3216446504 modified "2023-09-27" @default.
- W3216446504 title "Boosting Discriminative Visual Representation Learning with Scenario-Agnostic Mixup" @default.
- W3216446504 cites W1797268635 @default.
- W3216446504 cites W1846799578 @default.
- W3216446504 cites W1861492603 @default.
- W3216446504 cites W2031489346 @default.
- W3216446504 cites W2117539524 @default.
- W3216446504 cites W2163605009 @default.
- W3216446504 cites W2163922914 @default.
- W3216446504 cites W2194775991 @default.
- W3216446504 cites W2518108298 @default.
- W3216446504 cites W2549139847 @default.
- W3216446504 cites W2741430497 @default.
- W3216446504 cites W2765407302 @default.
- W3216446504 cites W2797977484 @default.
- W3216446504 cites W2842511635 @default.
- W3216446504 cites W2883725317 @default.
- W3216446504 cites W2921861056 @default.
- W3216446504 cites W2963091558 @default.
- W3216446504 cites W2963150697 @default.
- W3216446504 cites W2963341956 @default.
- W3216446504 cites W2964048159 @default.
- W3216446504 cites W2970971581 @default.
- W3216446504 cites W2992308087 @default.
- W3216446504 cites W2994931756 @default.
- W3216446504 cites W2998835636 @default.
- W3216446504 cites W3009561768 @default.
- W3216446504 cites W3012410440 @default.
- W3216446504 cites W3034302069 @default.
- W3216446504 cites W3034576826 @default.
- W3216446504 cites W3034978746 @default.
- W3216446504 cites W3035224233 @default.
- W3216446504 cites W3035524453 @default.
- W3216446504 cites W3037204200 @default.
- W3216446504 cites W3095121901 @default.
- W3216446504 cites W3100345210 @default.
- W3216446504 cites W3101821705 @default.
- W3216446504 cites W3106428938 @default.
- W3216446504 cites W3106485021 @default.
- W3216446504 cites W3108632511 @default.
- W3216446504 cites W3108655343 @default.
- W3216446504 cites W3114400232 @default.
- W3216446504 cites W3118608800 @default.
- W3216446504 cites W3123939835 @default.
- W3216446504 cites W3126816608 @default.
- W3216446504 cites W3127819337 @default.
- W3216446504 cites W3128825919 @default.
- W3216446504 cites W3137863028 @default.
- W3216446504 cites W3139278653 @default.
- W3216446504 cites W3167204053 @default.
- W3216446504 cites W3189092450 @default.
- W3216446504 cites W639708223 @default.
- W3216446504 doi "https://doi.org/10.48550/arxiv.2111.15454" @default.
- W3216446504 hasPublicationYear "2021" @default.
- W3216446504 type Work @default.
- W3216446504 sameAs 3216446504 @default.
- W3216446504 citedByCount "1" @default.
- W3216446504 countsByYear W32164465042023 @default.
- W3216446504 crossrefType "posted-content" @default.
- W3216446504 hasAuthorship W3216446504A5011760791 @default.
- W3216446504 hasAuthorship W3216446504A5033050983 @default.
- W3216446504 hasAuthorship W3216446504A5048295582 @default.
- W3216446504 hasAuthorship W3216446504A5055640195 @default.
- W3216446504 hasAuthorship W3216446504A5082786719 @default.
- W3216446504 hasBestOaLocation W32164465041 @default.
- W3216446504 hasConcept C105795698 @default.
- W3216446504 hasConcept C119857082 @default.
- W3216446504 hasConcept C14036430 @default.
- W3216446504 hasConcept C153180895 @default.
- W3216446504 hasConcept C154945302 @default.
- W3216446504 hasConcept C162324750 @default.
- W3216446504 hasConcept C17744445 @default.
- W3216446504 hasConcept C187736073 @default.
- W3216446504 hasConcept C199539241 @default.
- W3216446504 hasConcept C27158222 @default.
- W3216446504 hasConcept C2776359362 @default.
- W3216446504 hasConcept C2780451532 @default.
- W3216446504 hasConcept C33923547 @default.
- W3216446504 hasConcept C41008148 @default.
- W3216446504 hasConcept C46686674 @default.
- W3216446504 hasConcept C51632099 @default.
- W3216446504 hasConcept C78458016 @default.
- W3216446504 hasConcept C86803240 @default.
- W3216446504 hasConcept C94625758 @default.
- W3216446504 hasConcept C97931131 @default.
- W3216446504 hasConceptScore W3216446504C105795698 @default.
- W3216446504 hasConceptScore W3216446504C119857082 @default.
- W3216446504 hasConceptScore W3216446504C14036430 @default.
- W3216446504 hasConceptScore W3216446504C153180895 @default.
- W3216446504 hasConceptScore W3216446504C154945302 @default.
- W3216446504 hasConceptScore W3216446504C162324750 @default.
- W3216446504 hasConceptScore W3216446504C17744445 @default.