On Transversality of Bent Hyperplane Arrangements and the Topological Expressiveness of ReLU Neural Networks

DOI10.1137/20M1368902MaRDI QIDQ5073711zbMATH OpenFDO

Authors Elisenda Grigsby, Kathryn Lindsey

Publication date 3 May 2022

Published in SIAM Journal on Applied Algebra and Geometry (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/2008.09052

zbMATH Keywords

polyhedral complexesReLU neural networkstopological expressiveness

Mathematics Subject Classification ID

Polyhedral manifolds (52B70) Arrangements of points, flats, hyperplanes (aspects of discrete geometry) (52C35) PL-topology (57Q99) Critical points and critical submanifolds in differential topology (57R70) Artificial neural networks and deep learning (68T07)

Recommendations

• On decision regions of narrow deep neural networks
• On minimal representations of shallow ReLU networks
• Topology of deep neural networks
• Towards Lower Bounds on the Depth of ReLU Neural Networks
• Facets of neural network complexity

Cites work

• An introduction to hyperplane arrangements
• Approximation by superpositions of a sigmoidal function
• Convex Polytopes
• Facing up to arrangements: face-count formulas for partitions of space by hyperplanes
• scientific article; zbMATH DE number 5788282 (Why is no real title available?)
• Lectures on Polytopes
• Lower Bounds for Approximation by Nonlinear Manifolds
• On decision regions of narrow deep neural networks
• Reconciling modern machine-learning practice and the classical bias-variance trade-off

Cited in

(10)

• Some geometrical and topological properties of DNNs' decision boundaries
• Topology of deep neural networks
• Sharp Bounds for the Number of Regions of Maxout Networks and Vertices of Minkowski Sums
• On transversality of bent hyperplane arrangements and the topological expressiveness of ReLU neural networks
• Labeled sample compression schemes for complexes of oriented matroids
• Topological properties of basins of attraction of width bounded autoencoders
• Functional dimension of feedforward ReLU neural networks
• Convexity in ReLU neural networks: beyond ICNNs?
• Algorithmic determination of the combinatorial structure of the linear regions of ReLU neural networks
• Combinatorial regularity for relatively perfect discrete Morse gradient vector fields of ReLU neural networks

This page was built for publication: On Transversality of Bent Hyperplane Arrangements and the Topological Expressiveness of ReLU Neural Networks

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5073711)