FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3124193156> ?p ?o ?g. }

• W3124193156 abstract "Rehearsal is a critical component for class-incremental continual learning, yet it requires a substantial memory budget. Our work investigates whether we can significantly reduce this memory budget by leveraging unlabeled data from an agent's environment in a realistic and challenging continual learning paradigm. Specifically, we explore and formalize a novel semi-supervised continual learning (SSCL) setting, where labeled data is scarce yet non-i.i.d. unlabeled data from the agent's environment is plentiful. Importantly, data distributions in the SSCL setting are realistic and therefore reflect object class correlations between, and among, the labeled and unlabeled data distributions. We show that a strategy built on pseudo-labeling, consistency regularization, Out-of-Distribution (OoD) detection, and knowledge distillation reduces forgetting in this setting. Our approach, DistillMatch, increases performance over the state-of-the-art by no less than 8.7% average task accuracy and up to 54.5% average task accuracy in SSCL CIFAR-100 experiments. Moreover, we demonstrate that DistillMatch can save up to 0.23 stored images per processed unlabeled image compared to the next best method which only saves 0.08. Our results suggest that focusing on realistic correlated distributions is a significantly new perspective, which accentuates the importance of leveraging the world's structure as a continual learning strategy." @default.
• W3124193156 created "2021-02-01" @default.
• W3124193156 creator A5038819429 @default.
• W3124193156 creator A5040051663 @default.
• W3124193156 creator A5077808473 @default.
• W3124193156 creator A5088892134 @default.
• W3124193156 date "2021-01-23" @default.
• W3124193156 modified "2023-10-17" @default.
• W3124193156 title "Memory-Efficient Semi-Supervised Continual Learning: The World is its Own Replay Buffer" @default.
• W3124193156 cites W2108501770 @default.
• W3124193156 cites W2116522068 @default.
• W3124193156 cites W2178768799 @default.
• W3124193156 cites W2194775991 @default.
• W3124193156 cites W2473930607 @default.
• W3124193156 cites W2560647685 @default.
• W3124193156 cites W2592691248 @default.
• W3124193156 cites W2737492962 @default.
• W3124193156 cites W2753112586 @default.
• W3124193156 cites W2756815061 @default.
• W3124193156 cites W2765895026 @default.
• W3124193156 cites W2767414122 @default.
• W3124193156 cites W2788388592 @default.
• W3124193156 cites W2867167548 @default.
• W3124193156 cites W2884282566 @default.
• W3124193156 cites W2895723011 @default.
• W3124193156 cites W2899063268 @default.
• W3124193156 cites W2902625698 @default.
• W3124193156 cites W2926477959 @default.
• W3124193156 cites W2939137134 @default.
• W3124193156 cites W2946894912 @default.
• W3124193156 cites W2947461406 @default.
• W3124193156 cites W2949416428 @default.
• W3124193156 cites W2949777502 @default.
• W3124193156 cites W2950184636 @default.
• W3124193156 cites W2962724315 @default.
• W3124193156 cites W2962884963 @default.
• W3124193156 cites W2963038864 @default.
• W3124193156 cites W2963559848 @default.
• W3124193156 cites W2963560049 @default.
• W3124193156 cites W2963588172 @default.
• W3124193156 cites W2964151081 @default.
• W3124193156 cites W2964159205 @default.
• W3124193156 cites W2964189064 @default.
• W3124193156 cites W2968596670 @default.
• W3124193156 cites W2970066656 @default.
• W3124193156 cites W2970505118 @default.
• W3124193156 cites W2970586779 @default.
• W3124193156 cites W2971176100 @default.
• W3124193156 cites W2971524484 @default.
• W3124193156 cites W2975176181 @default.
• W3124193156 cites W2978426779 @default.
• W3124193156 cites W2982220706 @default.
• W3124193156 cites W2993466051 @default.
• W3124193156 cites W2995795252 @default.
• W3124193156 cites W2996653965 @default.
• W3124193156 cites W2997946860 @default.
• W3124193156 cites W3001197829 @default.
• W3124193156 cites W3002701914 @default.
• W3124193156 cites W3008449794 @default.
• W3124193156 cites W3012735052 @default.
• W3124193156 cites W3017080957 @default.
• W3124193156 cites W3023371261 @default.
• W3124193156 cites W3034230713 @default.
• W3124193156 cites W3082456245 @default.
• W3124193156 cites W3118608800 @default.
• W3124193156 cites W2426267443 @default.
• W3124193156 cites W2551176409 @default.
• W3124193156 doi "https://doi.org/10.48550/arxiv.2101.09536" @default.
• W3124193156 hasPublicationYear "2021" @default.
• W3124193156 type Work @default.
• W3124193156 sameAs 3124193156 @default.
• W3124193156 citedByCount "2" @default.
• W3124193156 countsByYear W31241931562021 @default.
• W3124193156 crossrefType "posted-content" @default.
• W3124193156 hasAuthorship W3124193156A5038819429 @default.
• W3124193156 hasAuthorship W3124193156A5040051663 @default.
• W3124193156 hasAuthorship W3124193156A5077808473 @default.
• W3124193156 hasAuthorship W3124193156A5088892134 @default.
• W3124193156 hasBestOaLocation W31241931561 @default.
• W3124193156 hasConcept C111919701 @default.
• W3124193156 hasConcept C119857082 @default.
• W3124193156 hasConcept C138885662 @default.
• W3124193156 hasConcept C154945302 @default.
• W3124193156 hasConcept C162324750 @default.
• W3124193156 hasConcept C187736073 @default.
• W3124193156 hasConcept C2776135515 @default.
• W3124193156 hasConcept C2776145971 @default.
• W3124193156 hasConcept C2776436953 @default.
• W3124193156 hasConcept C2779960059 @default.
• W3124193156 hasConcept C2780451532 @default.
• W3124193156 hasConcept C41008148 @default.
• W3124193156 hasConcept C41895202 @default.
• W3124193156 hasConcept C7149132 @default.
• W3124193156 hasConceptScore W3124193156C111919701 @default.
• W3124193156 hasConceptScore W3124193156C119857082 @default.
• W3124193156 hasConceptScore W3124193156C138885662 @default.
• W3124193156 hasConceptScore W3124193156C154945302 @default.
• W3124193156 hasConceptScore W3124193156C162324750 @default.
• W3124193156 hasConceptScore W3124193156C187736073 @default.
• W3124193156 hasConceptScore W3124193156C2776135515 @default.

• next