Rational approximation techniques for analysis of neural networks

DOI10.1109/18.312168MaRDI QIDQ4308779zbMATH OpenOpenAlexFDO

Authors Kai-Yeung Siu, Vwani Roychowdhury, T. Kailath

Publication date 4 May 1995

Published in IEEE Transactions on Information Theory (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1109/18.312168

zbMATH Keywords

Boolean functionsharmonic analysismultilayer neural networksneural networksparityrational approximationsigmoidal networksthreshold circuits

Mathematics Subject Classification ID

Approximation by rational functions (41A20) Approximation by other special function classes (41A30) Numerical methods in Fourier analysis (65T99) Analytic circuit theory (94C05)

Recommendations

• Approximating threshold circuits by rational functions
• Bounds for the Computational Power and Learning Complexity of Analog Neural Nets
• scientific article; zbMATH DE number 852149
• Approximation and estimation bounds for artificial neural networks
• scientific article; zbMATH DE number 1843047

Cited in

(17)

• Negative results for approximation using single layer and multilayer feedforward neural networks
• A linear relation between input and first layer in neural networks
• Neural networks and Boolean functions
• Size and Energy of Threshold Circuits Computing Mod Functions
• A small decrease in the degree of a polynomial with a given sign function can exponentially increase its weight and length
• Energy and depth of threshold circuits
• Bounds for the Computational Power and Learning Complexity of Analog Neural Nets
• Breaking the Minsky--Papert Barrier for Constant-Depth Circuits
• The power of asymmetry in constant-depth circuits
• Average-case lower bounds and satisfiability algorithms for small threshold circuits
• General-Purpose Computation with Neural Networks: A Survey of Complexity Theoretic Results
• scientific article; zbMATH DE number 773997 (Why is no real title available?)
• Near-Optimal Lower Bounds on the Threshold Degree and Sign-Rank of AC^0
• Algorithmic Polynomials
• Signalling techniques and their effect on neural network implementation sizes
• Size-energy tradeoffs for unate circuits computing symmetric Boolean functions
• A size-depth trade-off for the analog computation of Boolean functions

This page was built for publication: Rational approximation techniques for analysis of neural networks

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4308779)