Matches in SemOpenAlex for { <https://semopenalex.org/work/W3100852746> ?p ?o ?g. }
Showing items 1 to 98 of
98
with 100 items per page.
- W3100852746 endingPage "10646" @default.
- W3100852746 startingPage "10635" @default.
- W3100852746 abstract "Modifications to a neural network's input and output layers are often required to accommodate the specificities of most practical learning tasks. However, the impact of such changes on architecture's approximation capabilities is largely not understood. We present general conditions describing feature and readout maps that preserve an architecture's ability to approximate any continuous functions uniformly on compacts. As an application, we show that if an architecture is capable of universal approximation, then modifying its final layer to produce binary values creates a new architecture capable of deterministically approximating any classifier. In particular, we obtain guarantees for deep CNNs and deep feed-forward networks. Our results also have consequences within the scope of geometric deep learning. Specifically, when the input and output spaces are Cartan-Hadamard manifolds, we obtain geometrically meaningful feature and readout maps satisfying our criteria. Consequently, commonly used non-Euclidean regression models between spaces of symmetric positive definite matrices are extended to universal DNNs. The same result allows us to show that the hyperbolic feed-forward networks, used for hierarchical learning, are universal. Our result is also used to show that the common practice of randomizing all but the last two layers of a DNN produces a universal family of functions with probability one. We also provide conditions on a DNN's first (resp. last) few layer's connections and activation function which guarantee that these layers can have a width equal to the input (resp. output) space's dimension while not negatively affecting the architecture's approximation capabilities." @default.
- W3100852746 created "2020-11-23" @default.
- W3100852746 creator A5022434901 @default.
- W3100852746 creator A5036113771 @default.
- W3100852746 date "2020-06-03" @default.
- W3100852746 modified "2023-09-24" @default.
- W3100852746 title "Non-Euclidean Universal Approximation" @default.
- W3100852746 hasPublicationYear "2020" @default.
- W3100852746 type Work @default.
- W3100852746 sameAs 3100852746 @default.
- W3100852746 citedByCount "6" @default.
- W3100852746 countsByYear W31008527462020 @default.
- W3100852746 countsByYear W31008527462021 @default.
- W3100852746 crossrefType "proceedings-article" @default.
- W3100852746 hasAuthorship W3100852746A5022434901 @default.
- W3100852746 hasAuthorship W3100852746A5036113771 @default.
- W3100852746 hasConcept C108583219 @default.
- W3100852746 hasConcept C11413529 @default.
- W3100852746 hasConcept C123657996 @default.
- W3100852746 hasConcept C129782007 @default.
- W3100852746 hasConcept C138885662 @default.
- W3100852746 hasConcept C14036430 @default.
- W3100852746 hasConcept C142362112 @default.
- W3100852746 hasConcept C153349607 @default.
- W3100852746 hasConcept C154945302 @default.
- W3100852746 hasConcept C186450821 @default.
- W3100852746 hasConcept C202444582 @default.
- W3100852746 hasConcept C2524010 @default.
- W3100852746 hasConcept C2776401178 @default.
- W3100852746 hasConcept C2984842247 @default.
- W3100852746 hasConcept C33676613 @default.
- W3100852746 hasConcept C33923547 @default.
- W3100852746 hasConcept C41008148 @default.
- W3100852746 hasConcept C41895202 @default.
- W3100852746 hasConcept C48372109 @default.
- W3100852746 hasConcept C50644808 @default.
- W3100852746 hasConcept C78458016 @default.
- W3100852746 hasConcept C80444323 @default.
- W3100852746 hasConcept C86803240 @default.
- W3100852746 hasConcept C91873725 @default.
- W3100852746 hasConcept C94375191 @default.
- W3100852746 hasConcept C95623464 @default.
- W3100852746 hasConceptScore W3100852746C108583219 @default.
- W3100852746 hasConceptScore W3100852746C11413529 @default.
- W3100852746 hasConceptScore W3100852746C123657996 @default.
- W3100852746 hasConceptScore W3100852746C129782007 @default.
- W3100852746 hasConceptScore W3100852746C138885662 @default.
- W3100852746 hasConceptScore W3100852746C14036430 @default.
- W3100852746 hasConceptScore W3100852746C142362112 @default.
- W3100852746 hasConceptScore W3100852746C153349607 @default.
- W3100852746 hasConceptScore W3100852746C154945302 @default.
- W3100852746 hasConceptScore W3100852746C186450821 @default.
- W3100852746 hasConceptScore W3100852746C202444582 @default.
- W3100852746 hasConceptScore W3100852746C2524010 @default.
- W3100852746 hasConceptScore W3100852746C2776401178 @default.
- W3100852746 hasConceptScore W3100852746C2984842247 @default.
- W3100852746 hasConceptScore W3100852746C33676613 @default.
- W3100852746 hasConceptScore W3100852746C33923547 @default.
- W3100852746 hasConceptScore W3100852746C41008148 @default.
- W3100852746 hasConceptScore W3100852746C41895202 @default.
- W3100852746 hasConceptScore W3100852746C48372109 @default.
- W3100852746 hasConceptScore W3100852746C50644808 @default.
- W3100852746 hasConceptScore W3100852746C78458016 @default.
- W3100852746 hasConceptScore W3100852746C80444323 @default.
- W3100852746 hasConceptScore W3100852746C86803240 @default.
- W3100852746 hasConceptScore W3100852746C91873725 @default.
- W3100852746 hasConceptScore W3100852746C94375191 @default.
- W3100852746 hasConceptScore W3100852746C95623464 @default.
- W3100852746 hasLocation W31008527461 @default.
- W3100852746 hasOpenAccess W3100852746 @default.
- W3100852746 hasPrimaryLocation W31008527461 @default.
- W3100852746 hasRelatedWork W1480489032 @default.
- W3100852746 hasRelatedWork W1680674731 @default.
- W3100852746 hasRelatedWork W2103496339 @default.
- W3100852746 hasRelatedWork W2319967243 @default.
- W3100852746 hasRelatedWork W2894812324 @default.
- W3100852746 hasRelatedWork W2963918359 @default.
- W3100852746 hasRelatedWork W2969755609 @default.
- W3100852746 hasRelatedWork W3003298109 @default.
- W3100852746 hasRelatedWork W3007598458 @default.
- W3100852746 hasRelatedWork W3041921867 @default.
- W3100852746 hasRelatedWork W3049740931 @default.
- W3100852746 hasRelatedWork W3094996462 @default.
- W3100852746 hasRelatedWork W3111203218 @default.
- W3100852746 hasRelatedWork W3125537303 @default.
- W3100852746 hasRelatedWork W3126988966 @default.
- W3100852746 hasRelatedWork W3134120159 @default.
- W3100852746 hasRelatedWork W3137474564 @default.
- W3100852746 hasRelatedWork W3161375300 @default.
- W3100852746 hasRelatedWork W3186073754 @default.
- W3100852746 hasRelatedWork W2187753657 @default.
- W3100852746 hasVolume "33" @default.
- W3100852746 isParatext "false" @default.
- W3100852746 isRetracted "false" @default.
- W3100852746 magId "3100852746" @default.
- W3100852746 workType "article" @default.