Symmetry & Critical Points

Yossi Arjevani
8/26/2024
cs.LGmath.NAmath.OC

No confident code implementation found for this paper.

Abstract

Critical points of an invariant function may or may not be symmetric. We prove, however, that if a symmetric critical point exists, those adjacent to it are generically symmetry breaking. This mathematical mechanism is shown to carry important implications for our ability to efficiently minimize invariant nonconvex functions, in particular those associated with neural networks.

View on arXivView PDF

Code Implementations

No confident code match yet

We couldn't find an author-owned or strongly-evidenced community implementation for this paper. Any repos shown below are weak matches — verify before relying on them.

No code implementations found yet.

Know of an implementation? Let us know in the comments below!

Similar Papers

Ubiquitous Symmetry at Critical Points Across Diverse Optimization Landscapes

82% match2025

Symmetry & Critical Points for Symmetric Tensor Decomposition Problems

81% match2023

Symmetry Breaking and Equivariant Neural Networks

78% match2023

Parameter-Level Attribution of Symmetry in Trained Networks Though Parameter-Wise Functional Sensitivity

77% match2026

Remove Symmetries to Control Model Expressivity

76% match1 repo2024

Symmetry Breaking in Neural Network Optimization: Insights from Input Dimension Expansion

76% match2024

Cite this paper

@article{arjevani2024symmetry,
  title  = {Symmetry & Critical Points},
  author = {Yossi Arjevani},
  year   = {2024},
  eprint = {2408.14445},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2408.14445v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2408.14445.svg)](https://codeofpaper.com/paper/2408.14445)

Discussion