Matches in SemOpenAlex for { <https://semopenalex.org/work/W2970889419> ?p ?o ?g. }
Showing items 1 to 88 of
88
with 100 items per page.
- W2970889419 endingPage "9960" @default.
- W2970889419 startingPage "9947" @default.
- W2970889419 abstract "Wide neural networks with random weights and biases are Gaussian processes, as observed by Neal (1995) for shallow networks, and more recently by Lee et al.~(2018) and Matthews et al.~(2018) for deep fully-connected networks, as well as by Novak et al.~(2019) and Garriga-Alonso et al.~(2019) for deep convolutional networks. We show that this Neural Network-Gaussian Process correspondence surprisingly extends to all modern feedforward or recurrent neural networks composed of multilayer perceptron, RNNs (e.g. LSTMs, GRUs), (nD or graph) convolution, pooling, skip connection, attention, batch normalization, and/or layer normalization. More generally, we introduce a language for expressing neural network computations, and our result encompasses all such expressible neural networks. This work serves as a tutorial on the emph{tensor programs} technique formulated in Yang (2019) and elucidates the Gaussian Process results obtained there. We provide open-source implementations of the Gaussian Process kernels of simple RNN, GRU, transformer, and batchnorm+ReLU network at github.com/thegregyang/GP4A. Please see our arxiv version for the complete and up-to-date version of this paper." @default.
- W2970889419 created "2019-09-05" @default.
- W2970889419 creator A5061115157 @default.
- W2970889419 date "2019-12-08" @default.
- W2970889419 modified "2023-10-18" @default.
- W2970889419 title "Wide Feedforward or Recurrent Neural Networks of Any Architecture are Gaussian Processes" @default.
- W2970889419 hasPublicationYear "2019" @default.
- W2970889419 type Work @default.
- W2970889419 sameAs 2970889419 @default.
- W2970889419 citedByCount "28" @default.
- W2970889419 countsByYear W29708894192019 @default.
- W2970889419 countsByYear W29708894192020 @default.
- W2970889419 countsByYear W29708894192021 @default.
- W2970889419 countsByYear W29708894192022 @default.
- W2970889419 crossrefType "proceedings-article" @default.
- W2970889419 hasAuthorship W2970889419A5061115157 @default.
- W2970889419 hasConcept C108583219 @default.
- W2970889419 hasConcept C11413529 @default.
- W2970889419 hasConcept C121332964 @default.
- W2970889419 hasConcept C127413603 @default.
- W2970889419 hasConcept C133731056 @default.
- W2970889419 hasConcept C136886441 @default.
- W2970889419 hasConcept C144024400 @default.
- W2970889419 hasConcept C147168706 @default.
- W2970889419 hasConcept C154945302 @default.
- W2970889419 hasConcept C163716315 @default.
- W2970889419 hasConcept C19165224 @default.
- W2970889419 hasConcept C38858127 @default.
- W2970889419 hasConcept C41008148 @default.
- W2970889419 hasConcept C47702885 @default.
- W2970889419 hasConcept C50644808 @default.
- W2970889419 hasConcept C60908668 @default.
- W2970889419 hasConcept C61326573 @default.
- W2970889419 hasConcept C62520636 @default.
- W2970889419 hasConcept C80444323 @default.
- W2970889419 hasConcept C81363708 @default.
- W2970889419 hasConcept C86582703 @default.
- W2970889419 hasConceptScore W2970889419C108583219 @default.
- W2970889419 hasConceptScore W2970889419C11413529 @default.
- W2970889419 hasConceptScore W2970889419C121332964 @default.
- W2970889419 hasConceptScore W2970889419C127413603 @default.
- W2970889419 hasConceptScore W2970889419C133731056 @default.
- W2970889419 hasConceptScore W2970889419C136886441 @default.
- W2970889419 hasConceptScore W2970889419C144024400 @default.
- W2970889419 hasConceptScore W2970889419C147168706 @default.
- W2970889419 hasConceptScore W2970889419C154945302 @default.
- W2970889419 hasConceptScore W2970889419C163716315 @default.
- W2970889419 hasConceptScore W2970889419C19165224 @default.
- W2970889419 hasConceptScore W2970889419C38858127 @default.
- W2970889419 hasConceptScore W2970889419C41008148 @default.
- W2970889419 hasConceptScore W2970889419C47702885 @default.
- W2970889419 hasConceptScore W2970889419C50644808 @default.
- W2970889419 hasConceptScore W2970889419C60908668 @default.
- W2970889419 hasConceptScore W2970889419C61326573 @default.
- W2970889419 hasConceptScore W2970889419C62520636 @default.
- W2970889419 hasConceptScore W2970889419C80444323 @default.
- W2970889419 hasConceptScore W2970889419C81363708 @default.
- W2970889419 hasConceptScore W2970889419C86582703 @default.
- W2970889419 hasLocation W29708894191 @default.
- W2970889419 hasOpenAccess W2970889419 @default.
- W2970889419 hasPrimaryLocation W29708894191 @default.
- W2970889419 hasRelatedWork W1567512734 @default.
- W2970889419 hasRelatedWork W2056099894 @default.
- W2970889419 hasRelatedWork W2154087390 @default.
- W2970889419 hasRelatedWork W2167608136 @default.
- W2970889419 hasRelatedWork W2194775991 @default.
- W2970889419 hasRelatedWork W2809090039 @default.
- W2970889419 hasRelatedWork W2910655610 @default.
- W2970889419 hasRelatedWork W2913473169 @default.
- W2970889419 hasRelatedWork W2962685794 @default.
- W2970889419 hasRelatedWork W2962698540 @default.
- W2970889419 hasRelatedWork W2963323437 @default.
- W2970889419 hasRelatedWork W2964052793 @default.
- W2970889419 hasRelatedWork W2964088238 @default.
- W2970889419 hasRelatedWork W2970217468 @default.
- W2970889419 hasRelatedWork W2971043187 @default.
- W2970889419 hasRelatedWork W2994747787 @default.
- W2970889419 hasRelatedWork W3034979923 @default.
- W2970889419 hasRelatedWork W3101069636 @default.
- W2970889419 hasRelatedWork W3118608800 @default.
- W2970889419 hasRelatedWork W3180313059 @default.
- W2970889419 hasVolume "32" @default.
- W2970889419 isParatext "false" @default.
- W2970889419 isRetracted "false" @default.
- W2970889419 magId "2970889419" @default.
- W2970889419 workType "article" @default.