Absence of bottlenecks in a neural network determines its generic functional properties

DOI10.1134/S1064562420010160MaRDI QIDQ2243704zbMATH OpenOpenAlexFDO

Authors S. V. Kurochkin

Publication date 11 November 2021

Published in Doklady Mathematics (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1134/s1064562420010160

zbMATH Keywords

Morse functionneural network

Mathematics Subject Classification ID

Critical points and critical submanifolds in differential topology (57R70) Neural networks for/in biological studies, artificial life and related topics (92B20)

Recommendations

• Topological properties of the set of functions generated by neural networks of fixed size
• Reconstructing a neural net from its output
• Measure Theoretic Results for Approximation by Neural Networks with Limited Weights
• Neural network identifiability for a family of sigmoidal nonlinearities
• scientific article; zbMATH DE number 1225807

Cites work

• A note on optimization with Morse polynomials
• An invitation to Morse theory
• Approximation by superpositions of a sigmoidal function
• Elements of combinatorial and differential topology. Transl. from the Russian by Olga Sipacheva
• scientific article; zbMATH DE number 637070 (Why is no real title available?)
• scientific article; zbMATH DE number 1405266 (Why is no real title available?)
• scientific article; zbMATH DE number 2200453 (Why is no real title available?)
• Smooth functions statistics
• The matrix equation \(AX-YB=C\)
• Topology and data

Cited in

(3)

• Topological properties of the set of functions generated by neural networks of fixed size
• Neural network with smooth activation functions and without bottlenecks is almost surely a Morse function
• Affine symmetries and neural network identifiability

This page was built for publication: Absence of bottlenecks in a neural network determines its generic functional properties

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2243704)