Symmetry \& critical points for a model shallow neural network
From MaRDI portal
Publication:2077613
DOI10.1016/j.physd.2021.133014zbMath1491.37049arXiv2003.10576OpenAlexW3197486621MaRDI QIDQ2077613
Publication date: 21 February 2022
Published in: Physica D (Search for Journal in Brave)
Full work available at URL: https://arxiv.org/abs/2003.10576
critical pointssymmetry breakingpower series representationReLU activationspurious minimastudent-teacher network
Dynamical aspects of symmetries, equivariant bifurcation theory (37G40) Symmetries and invariants of dynamical systems (37C79)
Related Items (3)
On the Benefit of Width for Neural Networks: Disappearance of Basins ⋮ Unnamed Item ⋮ Equivariant bifurcation, quadratic equivariants, and symmetry breaking for the standard representation of S k
Uses Software
Cites Work
- Unnamed Item
- Unnamed Item
- Unnamed Item
- Unnamed Item
- Unnamed Item
- Unnamed Item
- Unnamed Item
- Unnamed Item
- Unnamed Item
- A classification of certain maximal subgroups of symmetric groups.
- Equivariant bifurcation theory and symmetry breaking
- A classification of the maximal subgroups of the finite alternating and symmetric groups
- Symmetry breaking and branching patterns in equivariant bifurcation theory. II
- Neural network with unbounded activation functions is universal approximator
- On the solutions of analytic equations
- Idéaux de fonctions différentiables. I
- Asynchronous networks and event driven dynamics
- Asynchronous networks: modularization of dynamics theorem
- Symmetry breaking in equivariant bifurcation problems
- Theoretical Insights Into the Optimization Landscape of Over-Parameterized Shallow Neural Networks
- Convergence Results for Neural Networks via Electrodynamics
- The classical Artin approximation theorems
- Dynamics of stochastic gradient descent for two-layer neural networks in the teacher–student setup*
- Maximal subgroups of finite groups
This page was built for publication: Symmetry \& critical points for a model shallow neural network
