FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2994872659> ?p ?o ?g. }

• W2994872659 abstract "Recent research shows that the following two models are equivalent: (a) infinitely wide neural networks (NNs) trained under l2 loss by gradient descent with infinitesimally small learning rate (b) kernel regression with respect to so-called Neural Tangent Kernels (NTKs) (Jacot et al., 2018). An efficient algorithm to compute the NTK, as well as its convolutional counterparts, appears in Arora et al. (2019a), which allowed studying performance of infinitely wide nets on datasets like CIFAR-10. However, super-quadratic running time of kernel methods makes them best suited for small-data tasks. We report results suggesting neural tangent kernels perform strongly on low-data tasks. 1. On a standard testbed of classification/regression tasks from the UCI database, NTK SVM beats the previous gold standard, Random Forests (RF), and also the corresponding finite nets. 2. On CIFAR-10 with 10 – 640 training samples, Convolutional NTK consistently beats ResNet-34 by 1% - 3%. 3. On VOC07 testbed for few-shot image classification tasks on ImageNet with transfer learning (Goyal et al., 2019), replacing the linear SVM currently used with a Convolutional NTK SVM consistently improves performance. 4. Comparing the performance of NTK with the finite-width net it was derived from, NTK behavior starts at lower net widths than suggested by theoretical analysis(Arora et al., 2019a). NTK’s efficacy may trace to lower variance of output." @default.
• W2994872659 created "2019-12-26" @default.
• W2994872659 creator A5013194814 @default.
• W2994872659 creator A5033061754 @default.
• W2994872659 creator A5034102301 @default.
• W2994872659 creator A5062744744 @default.
• W2994872659 creator A5071983998 @default.
• W2994872659 creator A5079951047 @default.
• W2994872659 date "2020-04-30" @default.
• W2994872659 modified "2023-09-23" @default.
• W2994872659 title "Harnessing the Power of Infinitely Wide Deep Nets on Small-data Tasks" @default.
• W2994872659 cites W1493832124 @default.
• W2994872659 cites W1560724230 @default.
• W2994872659 cites W2106525823 @default.
• W2994872659 cites W2108598243 @default.
• W2994872659 cites W2154087390 @default.
• W2994872659 cites W2167608136 @default.
• W2994872659 cites W2176412452 @default.
• W2994872659 cites W2194775991 @default.
• W2994872659 cites W2212676342 @default.
• W2994872659 cites W2558661413 @default.
• W2994872659 cites W2593958421 @default.
• W2994872659 cites W2785626633 @default.
• W2994872659 cites W2809090039 @default.
• W2994872659 cites W2886067286 @default.
• W2994872659 cites W2889798024 @default.
• W2994872659 cites W2894604724 @default.
• W2994872659 cites W2899748887 @default.
• W2994872659 cites W2900959181 @default.
• W2994872659 cites W2910655610 @default.
• W2994872659 cites W2911867426 @default.
• W2994872659 cites W2913473169 @default.
• W2994872659 cites W2954097612 @default.
• W2994872659 cites W2962698540 @default.
• W2994872659 cites W2962939986 @default.
• W2994872659 cites W2963323437 @default.
• W2994872659 cites W2964052793 @default.
• W2994872659 cites W2970239942 @default.
• W2994872659 cites W2970330753 @default.
• W2994872659 cites W2970474218 @default.
• W2994872659 cites W2971043187 @default.
• W2994872659 cites W2990873191 @default.
• W2994872659 cites W3137695714 @default.
• W2994872659 hasPublicationYear "2020" @default.
• W2994872659 type Work @default.
• W2994872659 sameAs 2994872659 @default.
• W2994872659 citedByCount "64" @default.
• W2994872659 countsByYear W29948726592019 @default.
• W2994872659 countsByYear W29948726592020 @default.
• W2994872659 countsByYear W29948726592021 @default.
• W2994872659 countsByYear W29948726592023 @default.
• W2994872659 crossrefType "proceedings-article" @default.
• W2994872659 hasAuthorship W2994872659A5013194814 @default.
• W2994872659 hasAuthorship W2994872659A5033061754 @default.
• W2994872659 hasAuthorship W2994872659A5034102301 @default.
• W2994872659 hasAuthorship W2994872659A5062744744 @default.
• W2994872659 hasAuthorship W2994872659A5071983998 @default.
• W2994872659 hasAuthorship W2994872659A5079951047 @default.
• W2994872659 hasConcept C11413529 @default.
• W2994872659 hasConcept C118615104 @default.
• W2994872659 hasConcept C12267149 @default.
• W2994872659 hasConcept C153180895 @default.
• W2994872659 hasConcept C153258448 @default.
• W2994872659 hasConcept C154945302 @default.
• W2994872659 hasConcept C33923547 @default.
• W2994872659 hasConcept C41008148 @default.
• W2994872659 hasConcept C50644808 @default.
• W2994872659 hasConcept C74193536 @default.
• W2994872659 hasConcept C81363708 @default.
• W2994872659 hasConceptScore W2994872659C11413529 @default.
• W2994872659 hasConceptScore W2994872659C118615104 @default.
• W2994872659 hasConceptScore W2994872659C12267149 @default.
• W2994872659 hasConceptScore W2994872659C153180895 @default.
• W2994872659 hasConceptScore W2994872659C153258448 @default.
• W2994872659 hasConceptScore W2994872659C154945302 @default.
• W2994872659 hasConceptScore W2994872659C33923547 @default.
• W2994872659 hasConceptScore W2994872659C41008148 @default.
• W2994872659 hasConceptScore W2994872659C50644808 @default.
• W2994872659 hasConceptScore W2994872659C74193536 @default.
• W2994872659 hasConceptScore W2994872659C81363708 @default.
• W2994872659 hasLocation W29948726591 @default.
• W2994872659 hasOpenAccess W2994872659 @default.
• W2994872659 hasPrimaryLocation W29948726591 @default.
• W2994872659 hasRelatedWork W2167608136 @default.
• W2994872659 hasRelatedWork W2194775991 @default.
• W2994872659 hasRelatedWork W2809090039 @default.
• W2994872659 hasRelatedWork W2899748887 @default.
• W2994872659 hasRelatedWork W2910655610 @default.
• W2994872659 hasRelatedWork W2913473169 @default.
• W2994872659 hasRelatedWork W2952204734 @default.
• W2994872659 hasRelatedWork W2962685794 @default.
• W2994872659 hasRelatedWork W2962698540 @default.
• W2994872659 hasRelatedWork W2962939986 @default.
• W2994872659 hasRelatedWork W2963239103 @default.
• W2994872659 hasRelatedWork W2963323437 @default.
• W2994872659 hasRelatedWork W2964052793 @default.
• W2994872659 hasRelatedWork W2970217468 @default.
• W2994872659 hasRelatedWork W2970330753 @default.
• W2994872659 hasRelatedWork W2970474218 @default.
• W2994872659 hasRelatedWork W2971043187 @default.

• next