Neural network identifiability for a family of sigmoidal nonlinearities

DOI10.1007/s00365-021-09544-3zbMath1482.68217arXiv1906.06994OpenAlexW3163739840MaRDI QIDQ2117333

Helmut Bölcskei, Verner Vlačić

Publication date: 21 March 2022

Published in: Constructive Approximation (Search for Journal in Brave)

Full work available at URL: https://arxiv.org/abs/1906.06994

zbMATH Keywords

identifiabilitydeep neural networkssigmoidal nonlinearities

Mathematics Subject Classification ID

Artificial neural networks and deep learning (68T07) Approximation by other special function classes (41A30)

Related Items (3)

Metric entropy limits on recurrent neural network learning of linear dynamical systems ⋮ Approximation error for neural network operators by an averaged modulus of smoothness ⋮ Stable recovery of entangled weights: towards robust identification of deep neural networks from minimal samples

Uses Software

• ImageNet
• AlexNet

Cites Work

• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Reconstructing a neural net from its output
• Multilayer feedforward networks are universal approximators
• Optimal approximation of piecewise smooth functions using deep ReLU neural networks
• Group Invariant Scattering
• A Mathematical Theory of Deep Convolutional Neural Networks for Feature Extraction
• A Course in Mathematical Analysis
• Optimal Approximation with Sparsely Connected Deep Neural Networks
• Lie Groups, Lie Algebras, and Representations
• Approximation by superpositions of a sigmoidal function

This page was built for publication: Neural network identifiability for a family of sigmoidal nonlinearities