Neural networks with linear threshold activations: structure and algorithms

DOI10.1007/S10107-023-02016-5MaRDI QIDQ6589753zbMATH OpenFDO

Authors Sammy Khalife, Hongyu Cheng, Amitabh Basu

Publication date 20 August 2024

Published in Mathematical Programming. Series A. Series B (Search for Journal in Brave)

zbMATH Keywords

neural networkspolyhedral theorycomplexity bounds

Mathematics Subject Classification ID

Artificial neural networks and deep learning (68T07) Polyhedral combinatorics, branch-and-bound, branch-and-cut (90C57) Analysis of algorithms and problem complexity (68Q25) Neural networks for/in biological studies, artificial life and related topics (92B20)

Recommendations

• Neural networks with linear threshold activations: structure and algorithms
• scientific article; zbMATH DE number 7250146
• Towards Lower Bounds on the Depth of ReLU Neural Networks
• Best \(k\)-layer neural network approximations
• Facets of neural network complexity

Cites work

• scientific article; zbMATH DE number 3385535 (Why is no real title available?)
• Approximating threshold circuits by rational functions
• Approximation Algorithms for Training One-Node ReLU Neural Networks
• Approximation by superpositions of a sigmoidal function
• Complexity of training ReLU neural network
• Neural Network Learning
• Robust trainability of single neurons
• Size--Depth Tradeoffs for Threshold Circuits
• Super-linear gate and super-quadratic wire lower bounds for depth-two and depth-three threshold circuits
• The Computational Complexity of ReLU Network Training Parameterized by Data Dimensionality

This page was built for publication: Neural networks with linear threshold activations: structure and algorithms

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6589753)