FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3046996896> ?p ?o ?g. }

• W3046996896 abstract "Recent work has attempted to interpret residual networks (ResNets) as one step of a forward Euler discretization of an ordinary differential equation, focusing mainly on syntactic algebraic similarities between the two systems. Discrete dynamical integrators of continuous dynamical systems, however, have a much richer structure. We first show that ResNets fail to be meaningful dynamical integrators in this richer sense. We then demonstrate that neural network models can learn to represent continuous dynamical systems, with this richer structure and properties, by embedding them into higher-order numerical integration schemes, such as the Runge Kutta schemes. Based on these insights, we introduce ContinuousNet as a continuous-in-depth generalization of ResNet architectures. ContinuousNets exhibit an invariance to the particular computational graph manifestation. That is, the continuous-in-depth model can be evaluated with different discrete time step sizes, which changes the number of layers, and different numerical integration schemes, which changes the graph connectivity. We show that this can be used to develop an incremental-in-depth training scheme that improves model quality, while significantly decreasing training time. We also show that, once trained, the number of units in the computational graph can even be decreased, for faster inference with little-to-no accuracy drop." @default.
• W3046996896 created "2020-08-10" @default.
• W3046996896 creator A5007032334 @default.
• W3046996896 creator A5025354752 @default.
• W3046996896 creator A5033006662 @default.
• W3046996896 creator A5037020807 @default.
• W3046996896 date "2020-08-05" @default.
• W3046996896 modified "2023-09-27" @default.
• W3046996896 title "Continuous-in-Depth Neural Networks." @default.
• W3046996896 cites W104184427 @default.
• W3046996896 cites W1519422254 @default.
• W3046996896 cites W1522301498 @default.
• W3046996896 cites W1677182931 @default.
• W3046996896 cites W1976846870 @default.
• W3046996896 cites W2025961705 @default.
• W3046996896 cites W2100677568 @default.
• W3046996896 cites W2118426468 @default.
• W3046996896 cites W2148101780 @default.
• W3046996896 cites W2194775991 @default.
• W3046996896 cites W2302255633 @default.
• W3046996896 cites W2401231614 @default.
• W3046996896 cites W2408279554 @default.
• W3046996896 cites W2528062157 @default.
• W3046996896 cites W2549139847 @default.
• W3046996896 cites W2591954064 @default.
• W3046996896 cites W2600297185 @default.
• W3046996896 cites W2766839578 @default.
• W3046996896 cites W2895434480 @default.
• W3046996896 cites W2899297079 @default.
• W3046996896 cites W2899771611 @default.
• W3046996896 cites W2914623577 @default.
• W3046996896 cites W2918156971 @default.
• W3046996896 cites W2930930439 @default.
• W3046996896 cites W2945102878 @default.
• W3046996896 cites W2947005146 @default.
• W3046996896 cites W2948977392 @default.
• W3046996896 cites W2952574409 @default.
• W3046996896 cites W2954591141 @default.
• W3046996896 cites W2962737770 @default.
• W3046996896 cites W2962933129 @default.
• W3046996896 cites W2963004026 @default.
• W3046996896 cites W2963021886 @default.
• W3046996896 cites W2963063862 @default.
• W3046996896 cites W2963172539 @default.
• W3046996896 cites W2963302407 @default.
• W3046996896 cites W2963359731 @default.
• W3046996896 cites W2963398562 @default.
• W3046996896 cites W2963410064 @default.
• W3046996896 cites W2963446712 @default.
• W3046996896 cites W2963456073 @default.
• W3046996896 cites W2963519230 @default.
• W3046996896 cites W2963540976 @default.
• W3046996896 cites W2963656894 @default.
• W3046996896 cites W2963684275 @default.
• W3046996896 cites W2963755523 @default.
• W3046996896 cites W2964088127 @default.
• W3046996896 cites W2964236304 @default.
• W3046996896 cites W2964627548 @default.
• W3046996896 cites W2970181183 @default.
• W3046996896 cites W2970247852 @default.
• W3046996896 cites W2970535233 @default.
• W3046996896 cites W2970701213 @default.
• W3046996896 cites W2970900903 @default.
• W3046996896 cites W2971086474 @default.
• W3046996896 cites W2971162686 @default.
• W3046996896 cites W2971343405 @default.
• W3046996896 cites W2974916071 @default.
• W3046996896 cites W2975784920 @default.
• W3046996896 cites W2977701733 @default.
• W3046996896 cites W2979563282 @default.
• W3046996896 cites W2981407965 @default.
• W3046996896 cites W2982479999 @default.
• W3046996896 cites W2986385482 @default.
• W3046996896 cites W2986615800 @default.
• W3046996896 cites W2986677756 @default.
• W3046996896 cites W2990121169 @default.
• W3046996896 cites W2994942528 @default.
• W3046996896 cites W2996160951 @default.
• W3046996896 cites W2996390082 @default.
• W3046996896 cites W2996487136 @default.
• W3046996896 cites W2996897098 @default.
• W3046996896 cites W2999026783 @default.
• W3046996896 cites W3005235017 @default.
• W3046996896 cites W3007175350 @default.
• W3046996896 cites W3007689282 @default.
• W3046996896 cites W3008349919 @default.
• W3046996896 cites W3009143787 @default.
• W3046996896 cites W3011162163 @default.
• W3046996896 cites W3021457401 @default.
• W3046996896 cites W3027429260 @default.
• W3046996896 cites W3027495034 @default.
• W3046996896 cites W3035794324 @default.
• W3046996896 cites W3035804453 @default.
• W3046996896 cites W3035985817 @default.
• W3046996896 cites W3098011980 @default.
• W3046996896 cites W3118608800 @default.
• W3046996896 cites W3034652062 @default.
• W3046996896 hasPublicationYear "2020" @default.
• W3046996896 type Work @default.
• W3046996896 sameAs 3046996896 @default.

• next