Matches in SemOpenAlex for { <https://semopenalex.org/work/W3203205487> ?p ?o ?g. }
- W3203205487 abstract "We consider iterative semi-supervised learning (SSL) algorithms that iteratively generate pseudo-labels for a large amount unlabelled data to progressively refine the model parameters. In particular, we seek to understand the behaviour of the {em generalization error} of iterative SSL algorithms using information-theoretic principles. To obtain bounds that are amenable to numerical evaluation, we first work with a simple model -- namely, the binary Gaussian mixture model. Our theoretical results suggest that when the class conditional variances are not too large, the upper bound on the generalization error decreases monotonically with the number of iterations, but quickly saturates. The theoretical results on the simple model are corroborated by extensive experiments on several benchmark datasets such as the MNIST and CIFAR datasets in which we notice that the generalization error improves after several pseudo-labelling iterations, but saturates afterwards." @default.
- W3203205487 created "2021-10-11" @default.
- W3203205487 creator A5058345431 @default.
- W3203205487 creator A5071743450 @default.
- W3203205487 creator A5073481627 @default.
- W3203205487 date "2021-10-03" @default.
- W3203205487 modified "2023-09-27" @default.
- W3203205487 title "Information-Theoretic Generalization Bounds for Iterative Semi-Supervised Learning." @default.
- W3203205487 cites W1479807131 @default.
- W3203205487 cites W1990334093 @default.
- W3203205487 cites W2014902932 @default.
- W3203205487 cites W2106401878 @default.
- W3203205487 cites W2112796928 @default.
- W3203205487 cites W2114771311 @default.
- W3203205487 cites W2136504847 @default.
- W3203205487 cites W2156909104 @default.
- W3203205487 cites W2166944917 @default.
- W3203205487 cites W2194775991 @default.
- W3203205487 cites W2258658829 @default.
- W3203205487 cites W2617539226 @default.
- W3203205487 cites W2910572885 @default.
- W3203205487 cites W2943889198 @default.
- W3203205487 cites W2962702650 @default.
- W3203205487 cites W2964121744 @default.
- W3203205487 cites W2967655751 @default.
- W3203205487 cites W2970638103 @default.
- W3203205487 cites W2970680991 @default.
- W3203205487 cites W2970941190 @default.
- W3203205487 cites W3037428746 @default.
- W3203205487 cites W3046489903 @default.
- W3203205487 cites W3093459558 @default.
- W3203205487 cites W3099022131 @default.
- W3203205487 cites W3100231902 @default.
- W3203205487 cites W3118608800 @default.
- W3203205487 cites W3159409174 @default.
- W3203205487 cites W3170524081 @default.
- W3203205487 cites W568673721 @default.
- W3203205487 cites W740415 @default.
- W3203205487 hasPublicationYear "2021" @default.
- W3203205487 type Work @default.
- W3203205487 sameAs 3203205487 @default.
- W3203205487 citedByCount "0" @default.
- W3203205487 crossrefType "posted-content" @default.
- W3203205487 hasAuthorship W3203205487A5058345431 @default.
- W3203205487 hasAuthorship W3203205487A5071743450 @default.
- W3203205487 hasAuthorship W3203205487A5073481627 @default.
- W3203205487 hasConcept C111472728 @default.
- W3203205487 hasConcept C11413529 @default.
- W3203205487 hasConcept C117619785 @default.
- W3203205487 hasConcept C117765406 @default.
- W3203205487 hasConcept C121332964 @default.
- W3203205487 hasConcept C13280743 @default.
- W3203205487 hasConcept C134306372 @default.
- W3203205487 hasConcept C138885662 @default.
- W3203205487 hasConcept C154945302 @default.
- W3203205487 hasConcept C159694833 @default.
- W3203205487 hasConcept C163716315 @default.
- W3203205487 hasConcept C177148314 @default.
- W3203205487 hasConcept C185798385 @default.
- W3203205487 hasConcept C190502265 @default.
- W3203205487 hasConcept C205649164 @default.
- W3203205487 hasConcept C2775924081 @default.
- W3203205487 hasConcept C2777212361 @default.
- W3203205487 hasConcept C2780586882 @default.
- W3203205487 hasConcept C33923547 @default.
- W3203205487 hasConcept C41008148 @default.
- W3203205487 hasConcept C48372109 @default.
- W3203205487 hasConcept C50644808 @default.
- W3203205487 hasConcept C62520636 @default.
- W3203205487 hasConcept C94375191 @default.
- W3203205487 hasConceptScore W3203205487C111472728 @default.
- W3203205487 hasConceptScore W3203205487C11413529 @default.
- W3203205487 hasConceptScore W3203205487C117619785 @default.
- W3203205487 hasConceptScore W3203205487C117765406 @default.
- W3203205487 hasConceptScore W3203205487C121332964 @default.
- W3203205487 hasConceptScore W3203205487C13280743 @default.
- W3203205487 hasConceptScore W3203205487C134306372 @default.
- W3203205487 hasConceptScore W3203205487C138885662 @default.
- W3203205487 hasConceptScore W3203205487C154945302 @default.
- W3203205487 hasConceptScore W3203205487C159694833 @default.
- W3203205487 hasConceptScore W3203205487C163716315 @default.
- W3203205487 hasConceptScore W3203205487C177148314 @default.
- W3203205487 hasConceptScore W3203205487C185798385 @default.
- W3203205487 hasConceptScore W3203205487C190502265 @default.
- W3203205487 hasConceptScore W3203205487C205649164 @default.
- W3203205487 hasConceptScore W3203205487C2775924081 @default.
- W3203205487 hasConceptScore W3203205487C2777212361 @default.
- W3203205487 hasConceptScore W3203205487C2780586882 @default.
- W3203205487 hasConceptScore W3203205487C33923547 @default.
- W3203205487 hasConceptScore W3203205487C41008148 @default.
- W3203205487 hasConceptScore W3203205487C48372109 @default.
- W3203205487 hasConceptScore W3203205487C50644808 @default.
- W3203205487 hasConceptScore W3203205487C62520636 @default.
- W3203205487 hasConceptScore W3203205487C94375191 @default.
- W3203205487 hasLocation W32032054871 @default.
- W3203205487 hasOpenAccess W3203205487 @default.
- W3203205487 hasPrimaryLocation W32032054871 @default.
- W3203205487 hasRelatedWork W1560379207 @default.
- W3203205487 hasRelatedWork W1588148453 @default.
- W3203205487 hasRelatedWork W2140076625 @default.