Matches in SemOpenAlex for { <https://semopenalex.org/work/W2955957078> ?p ?o ?g. }
Showing items 1 to 85 of
85
with 100 items per page.
- W2955957078 abstract "In representation learning and non-linear dimension reduction, there is a huge interest to learn the 'disentangled' latent variables, where each sub-coordinate almost uniquely controls a facet of the observed data. While many regularization approaches have been proposed on variational autoencoders, heuristic tuning is required to balance between disentanglement and loss in reconstruction accuracy -- due to the unsupervised nature, there is no principled way to find an optimal weight for regularization. Motivated to completely bypass regularization, we consider a projection strategy: modifying the canonical Gaussian encoder, we add a layer of scaling and rotation to the Gaussian mean, such that the marginal correlations among latent sub-coordinates become exactly zero. This achieves a theoretically maximal disentanglement, as guaranteed by zero cross-correlation between one latent sub-coordinate and the observed varying with the rest. Unlike regularizations, the extra projection layer does not impact the flexibility of the previous encoder layers, leading to almost no loss in expressiveness. This approach is simple to implement in practice. Our numerical experiments demonstrate very good performance, with no tuning required." @default.
- W2955957078 created "2019-07-12" @default.
- W2955957078 creator A5002270240 @default.
- W2955957078 creator A5049615306 @default.
- W2955957078 date "2019-06-27" @default.
- W2955957078 modified "2023-09-24" @default.
- W2955957078 title "Tuning-Free Disentanglement via Projection" @default.
- W2955957078 cites W1834627138 @default.
- W2955957078 cites W1915328033 @default.
- W2955957078 cites W2010625607 @default.
- W2955957078 cites W2022139156 @default.
- W2955957078 cites W2047118428 @default.
- W2955957078 cites W2054640142 @default.
- W2955957078 cites W2162250452 @default.
- W2955957078 cites W2181083374 @default.
- W2955957078 cites W2753738274 @default.
- W2955957078 cites W2785519580 @default.
- W2955957078 cites W2902227449 @default.
- W2955957078 cites W2914441330 @default.
- W2955957078 cites W2951004968 @default.
- W2955957078 cites W2951873722 @default.
- W2955957078 cites W2963226019 @default.
- W2955957078 cites W2963366547 @default.
- W2955957078 doi "https://doi.org/10.48550/arxiv.1906.11732" @default.
- W2955957078 hasPublicationYear "2019" @default.
- W2955957078 type Work @default.
- W2955957078 sameAs 2955957078 @default.
- W2955957078 citedByCount "0" @default.
- W2955957078 crossrefType "posted-content" @default.
- W2955957078 hasAuthorship W2955957078A5002270240 @default.
- W2955957078 hasAuthorship W2955957078A5049615306 @default.
- W2955957078 hasBestOaLocation W29559570781 @default.
- W2955957078 hasConcept C105795698 @default.
- W2955957078 hasConcept C11413529 @default.
- W2955957078 hasConcept C114614502 @default.
- W2955957078 hasConcept C118505674 @default.
- W2955957078 hasConcept C121332964 @default.
- W2955957078 hasConcept C154945302 @default.
- W2955957078 hasConcept C163716315 @default.
- W2955957078 hasConcept C2524010 @default.
- W2955957078 hasConcept C2776135515 @default.
- W2955957078 hasConcept C28826006 @default.
- W2955957078 hasConcept C33676613 @default.
- W2955957078 hasConcept C33923547 @default.
- W2955957078 hasConcept C41008148 @default.
- W2955957078 hasConcept C51167844 @default.
- W2955957078 hasConcept C57493831 @default.
- W2955957078 hasConcept C62520636 @default.
- W2955957078 hasConcept C70518039 @default.
- W2955957078 hasConcept C99844830 @default.
- W2955957078 hasConceptScore W2955957078C105795698 @default.
- W2955957078 hasConceptScore W2955957078C11413529 @default.
- W2955957078 hasConceptScore W2955957078C114614502 @default.
- W2955957078 hasConceptScore W2955957078C118505674 @default.
- W2955957078 hasConceptScore W2955957078C121332964 @default.
- W2955957078 hasConceptScore W2955957078C154945302 @default.
- W2955957078 hasConceptScore W2955957078C163716315 @default.
- W2955957078 hasConceptScore W2955957078C2524010 @default.
- W2955957078 hasConceptScore W2955957078C2776135515 @default.
- W2955957078 hasConceptScore W2955957078C28826006 @default.
- W2955957078 hasConceptScore W2955957078C33676613 @default.
- W2955957078 hasConceptScore W2955957078C33923547 @default.
- W2955957078 hasConceptScore W2955957078C41008148 @default.
- W2955957078 hasConceptScore W2955957078C51167844 @default.
- W2955957078 hasConceptScore W2955957078C57493831 @default.
- W2955957078 hasConceptScore W2955957078C62520636 @default.
- W2955957078 hasConceptScore W2955957078C70518039 @default.
- W2955957078 hasConceptScore W2955957078C99844830 @default.
- W2955957078 hasLocation W29559570781 @default.
- W2955957078 hasOpenAccess W2955957078 @default.
- W2955957078 hasPrimaryLocation W29559570781 @default.
- W2955957078 hasRelatedWork W1484758315 @default.
- W2955957078 hasRelatedWork W2039896269 @default.
- W2955957078 hasRelatedWork W2361401480 @default.
- W2955957078 hasRelatedWork W2365203999 @default.
- W2955957078 hasRelatedWork W2367521487 @default.
- W2955957078 hasRelatedWork W2369293910 @default.
- W2955957078 hasRelatedWork W2377950740 @default.
- W2955957078 hasRelatedWork W2952130664 @default.
- W2955957078 hasRelatedWork W3104966664 @default.
- W2955957078 hasRelatedWork W2044757589 @default.
- W2955957078 isParatext "false" @default.
- W2955957078 isRetracted "false" @default.
- W2955957078 magId "2955957078" @default.
- W2955957078 workType "article" @default.