FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3078407515> ?p ?o ?g. }

• W3078407515 abstract "We investigate the expressive power of depth-2 bandlimited random neural networks. A random net is a neural network where the hidden layer parameters are frozen with random assignment, and only the output layer parameters are trained by loss minimization. Using random weights for a hidden layer is an effective method to avoid non-convex optimization in standard gradient descent learning. It has also been adopted in recent deep learning theories. Despite the well-known fact that a neural network is a universal approximator, in this study, we mathematically show that when hidden parameters are distributed in a bounded domain, the network may not achieve zero approximation error. In particular, we derive a new nontrivial approximation error lower bound. The proof utilizes the technique of ridgelet analysis, a harmonic analysis method designed for neural networks. This method is inspired by fundamental principles in classical signal processing, specifically the idea that signals with limited bandwidth may not always be able to perfectly recreate the original signal. We corroborate our theoretical results with various simulation studies, and generally, two main take-home messages are offered: (i) Not any distribution for selecting random weights is feasible to build a universal approximator; (ii) A suitable assignment of random weights exists but to some degree is associated with the complexity of the target function." @default.
• W3078407515 created "2020-08-24" @default.
• W3078407515 creator A5011259248 @default.
• W3078407515 creator A5021384155 @default.
• W3078407515 creator A5036537988 @default.
• W3078407515 creator A5045211553 @default.
• W3078407515 creator A5070236689 @default.
• W3078407515 date "2020-08-19" @default.
• W3078407515 modified "2023-09-23" @default.
• W3078407515 title "How Powerful are Shallow Neural Networks with Bandlimited Random Weights?" @default.
• W3078407515 cites W1567512734 @default.
• W3078407515 cites W1587245433 @default.
• W3078407515 cites W1613359937 @default.
• W3078407515 cites W1625958017 @default.
• W3078407515 cites W1786513448 @default.
• W3078407515 cites W184557537 @default.
• W3078407515 cites W1971735090 @default.
• W3078407515 cites W1977941534 @default.
• W3078407515 cites W1982192825 @default.
• W3078407515 cites W1986278072 @default.
• W3078407515 cites W1992282307 @default.
• W3078407515 cites W1996640396 @default.
• W3078407515 cites W2031299600 @default.
• W3078407515 cites W2032814004 @default.
• W3078407515 cites W2040870580 @default.
• W3078407515 cites W2089497633 @default.
• W3078407515 cites W2123395972 @default.
• W3078407515 cites W2144902422 @default.
• W3078407515 cites W2150354929 @default.
• W3078407515 cites W2155910151 @default.
• W3078407515 cites W2166116275 @default.
• W3078407515 cites W2167608136 @default.
• W3078407515 cites W2171865010 @default.
• W3078407515 cites W2528305538 @default.
• W3078407515 cites W2532798880 @default.
• W3078407515 cites W2560247988 @default.
• W3078407515 cites W2573520116 @default.
• W3078407515 cites W2586160710 @default.
• W3078407515 cites W2593382986 @default.
• W3078407515 cites W2593958421 @default.
• W3078407515 cites W2738226240 @default.
• W3078407515 cites W2752851182 @default.
• W3078407515 cites W2753648062 @default.
• W3078407515 cites W2798865883 @default.
• W3078407515 cites W2803636134 @default.
• W3078407515 cites W2809090039 @default.
• W3078407515 cites W2952204734 @default.
• W3078407515 cites W2952594495 @default.
• W3078407515 cites W2963518130 @default.
• W3078407515 cites W2963650649 @default.
• W3078407515 cites W2963709899 @default.
• W3078407515 cites W2964153674 @default.
• W3078407515 cites W2964624822 @default.
• W3078407515 cites W2970618525 @default.
• W3078407515 cites W2970723196 @default.
• W3078407515 cites W2979473749 @default.
• W3078407515 cites W2992906545 @default.
• W3078407515 cites W2995354826 @default.
• W3078407515 cites W2996492157 @default.
• W3078407515 cites W3004334800 @default.
• W3078407515 cites W3019297623 @default.
• W3078407515 cites W3034649595 @default.
• W3078407515 cites W3041739345 @default.
• W3078407515 cites W3097609957 @default.
• W3078407515 cites W3101806332 @default.
• W3078407515 cites W3101837697 @default.
• W3078407515 cites W3108216709 @default.
• W3078407515 cites W57510865 @default.
• W3078407515 cites W78356000 @default.
• W3078407515 cites W2180383608 @default.
• W3078407515 doi "https://doi.org/10.48550/arxiv.2008.08427" @default.
• W3078407515 hasPublicationYear "2020" @default.
• W3078407515 type Work @default.
• W3078407515 sameAs 3078407515 @default.
• W3078407515 citedByCount "1" @default.
• W3078407515 countsByYear W30784075152021 @default.
• W3078407515 crossrefType "posted-content" @default.
• W3078407515 hasAuthorship W3078407515A5011259248 @default.
• W3078407515 hasAuthorship W3078407515A5021384155 @default.
• W3078407515 hasAuthorship W3078407515A5036537988 @default.
• W3078407515 hasAuthorship W3078407515A5045211553 @default.
• W3078407515 hasAuthorship W3078407515A5070236689 @default.
• W3078407515 hasBestOaLocation W30784075151 @default.
• W3078407515 hasConcept C102519508 @default.
• W3078407515 hasConcept C11413529 @default.
• W3078407515 hasConcept C129997835 @default.
• W3078407515 hasConcept C134306372 @default.
• W3078407515 hasConcept C153258448 @default.
• W3078407515 hasConcept C154945302 @default.
• W3078407515 hasConcept C206688291 @default.
• W3078407515 hasConcept C33923547 @default.
• W3078407515 hasConcept C34388435 @default.
• W3078407515 hasConcept C41008148 @default.
• W3078407515 hasConcept C50644808 @default.
• W3078407515 hasConceptScore W3078407515C102519508 @default.
• W3078407515 hasConceptScore W3078407515C11413529 @default.
• W3078407515 hasConceptScore W3078407515C129997835 @default.
• W3078407515 hasConceptScore W3078407515C134306372 @default.
• W3078407515 hasConceptScore W3078407515C153258448 @default.
• W3078407515 hasConceptScore W3078407515C154945302 @default.

• next