The complexity of explaining neural networks through (group) invariants

DOI10.1016/J.TCS.2019.11.012MaRDI QIDQ2290685zbMATH OpenOpenAlexFDO

Authors Danielle Ensign, Scott Neville, Arnab Paul, Suresh Venkatasubramanian

Publication date 29 January 2020

Published in Theoretical Computer Science (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1016/j.tcs.2019.11.012

zbMATH Keywords

group theoryneural networksNP-completeness

Mathematics Subject Classification ID

Computational difficulty of problems (lower bounds, completeness, difficulty of approximation, etc.) (68Q17) Artificial neural networks and deep learning (68T07)

Recommendations

• The complexity of explaining neural networks through (group) invariants
• Probabilistic symmetries and invariant neural networks
• Representation theory and invariant neural networks
• A unifying framework for invariant pattern recognition
• Analysis of deep convolutional networks from group theory viewpoint

Cites work

• Approximation by superpositions of a sigmoidal function
• scientific article; zbMATH DE number 3314813 (Why is no real title available?)
• Mathematics of the neural response
• Multilayer feedforward networks are universal approximators

Cited in

(3)

• Constructing the basis path set by eliminating the path dependency
• The complexity of explaining neural networks through (group) invariants
• Probabilistic symmetries and invariant neural networks

This page was built for publication: The complexity of explaining neural networks through (group) invariants

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2290685)