Matches in SemOpenAlex for { <https://semopenalex.org/work/W4387074792> ?p ?o ?g. }
Showing items 1 to 75 of
75
with 100 items per page.
- W4387074792 abstract "A prominent goal of representation learning research is to achieve representations which are factorized in a useful manner with respect to the ground truth factors of variation. The fields of disentangled and equivariant representation learning have approached this ideal from a range of complimentary perspectives; however, to date, most approaches have proven to either be ill-specified or insufficiently flexible to effectively separate all realistic factors of interest in a learned latent space. In this work, we propose an alternative viewpoint on such structured representation learning which we call Flow Factorized Representation Learning, and demonstrate it to learn both more efficient and more usefully structured representations than existing frameworks. Specifically, we introduce a generative model which specifies a distinct set of latent probability paths that define different input transformations. Each latent flow is generated by the gradient field of a learned potential following dynamic optimal transport. Our novel setup brings new understandings to both textit{disentanglement} and textit{equivariance}. We show that our model achieves higher likelihoods on standard representation learning benchmarks while simultaneously being closer to approximately equivariant models. Furthermore, we demonstrate that the transformations learned by our model are flexibly composable and can also extrapolate to new data, implying a degree of robustness and generalizability approaching the ultimate goal of usefully factorized representation learning." @default.
- W4387074792 created "2023-09-27" @default.
- W4387074792 creator A5027171279 @default.
- W4387074792 creator A5037757022 @default.
- W4387074792 creator A5087368991 @default.
- W4387074792 creator A5088917149 @default.
- W4387074792 date "2023-09-22" @default.
- W4387074792 modified "2023-09-28" @default.
- W4387074792 title "Flow Factorized Representation Learning" @default.
- W4387074792 doi "https://doi.org/10.48550/arxiv.2309.13167" @default.
- W4387074792 hasPublicationYear "2023" @default.
- W4387074792 type Work @default.
- W4387074792 citedByCount "0" @default.
- W4387074792 crossrefType "posted-content" @default.
- W4387074792 hasAuthorship W4387074792A5027171279 @default.
- W4387074792 hasAuthorship W4387074792A5037757022 @default.
- W4387074792 hasAuthorship W4387074792A5087368991 @default.
- W4387074792 hasAuthorship W4387074792A5088917149 @default.
- W4387074792 hasBestOaLocation W43870747921 @default.
- W4387074792 hasConcept C104317684 @default.
- W4387074792 hasConcept C105795698 @default.
- W4387074792 hasConcept C119857082 @default.
- W4387074792 hasConcept C154945302 @default.
- W4387074792 hasConcept C167966045 @default.
- W4387074792 hasConcept C17744445 @default.
- W4387074792 hasConcept C185592680 @default.
- W4387074792 hasConcept C199539241 @default.
- W4387074792 hasConcept C202444582 @default.
- W4387074792 hasConcept C27158222 @default.
- W4387074792 hasConcept C2776359362 @default.
- W4387074792 hasConcept C33923547 @default.
- W4387074792 hasConcept C39890363 @default.
- W4387074792 hasConcept C41008148 @default.
- W4387074792 hasConcept C55493867 @default.
- W4387074792 hasConcept C59404180 @default.
- W4387074792 hasConcept C63479239 @default.
- W4387074792 hasConcept C80444323 @default.
- W4387074792 hasConcept C94625758 @default.
- W4387074792 hasConcept C9652623 @default.
- W4387074792 hasConceptScore W4387074792C104317684 @default.
- W4387074792 hasConceptScore W4387074792C105795698 @default.
- W4387074792 hasConceptScore W4387074792C119857082 @default.
- W4387074792 hasConceptScore W4387074792C154945302 @default.
- W4387074792 hasConceptScore W4387074792C167966045 @default.
- W4387074792 hasConceptScore W4387074792C17744445 @default.
- W4387074792 hasConceptScore W4387074792C185592680 @default.
- W4387074792 hasConceptScore W4387074792C199539241 @default.
- W4387074792 hasConceptScore W4387074792C202444582 @default.
- W4387074792 hasConceptScore W4387074792C27158222 @default.
- W4387074792 hasConceptScore W4387074792C2776359362 @default.
- W4387074792 hasConceptScore W4387074792C33923547 @default.
- W4387074792 hasConceptScore W4387074792C39890363 @default.
- W4387074792 hasConceptScore W4387074792C41008148 @default.
- W4387074792 hasConceptScore W4387074792C55493867 @default.
- W4387074792 hasConceptScore W4387074792C59404180 @default.
- W4387074792 hasConceptScore W4387074792C63479239 @default.
- W4387074792 hasConceptScore W4387074792C80444323 @default.
- W4387074792 hasConceptScore W4387074792C94625758 @default.
- W4387074792 hasConceptScore W4387074792C9652623 @default.
- W4387074792 hasLocation W43870747921 @default.
- W4387074792 hasOpenAccess W4387074792 @default.
- W4387074792 hasPrimaryLocation W43870747921 @default.
- W4387074792 hasRelatedWork W1534961803 @default.
- W4387074792 hasRelatedWork W2251149342 @default.
- W4387074792 hasRelatedWork W2335364074 @default.
- W4387074792 hasRelatedWork W3006036127 @default.
- W4387074792 hasRelatedWork W3035415268 @default.
- W4387074792 hasRelatedWork W3171895902 @default.
- W4387074792 hasRelatedWork W4200511449 @default.
- W4387074792 hasRelatedWork W4281766347 @default.
- W4387074792 hasRelatedWork W4287122539 @default.
- W4387074792 hasRelatedWork W4300480195 @default.
- W4387074792 isParatext "false" @default.
- W4387074792 isRetracted "false" @default.
- W4387074792 workType "article" @default.