Informative Fisher Modes in Neural Networks as a Result of Explicit Symmetry Breaking
cond-mat.dis-nn, cond-mat.stat-mech, math-ph, math.MP
Submitted: 2026-10-31
Updated: 2026-10-31
Project page: https://mathai-iclr.github.io/papers/papers/MATHAI_29_paper.pdf
Terminology
Sources
- Noise-Driven Escape from Metastable Phases explains Grokking in Deep Neural Networks
- Cascading Through the Hierarchy: Regularizer-Induced Feature Detection as Phase Transitions in Deep Linear Neural Networks
- Regularization Implies balancedness in the deep linear network
- Parameter Symmetry Potentially Unifies Deep Learning Theory
- Remove Symmetries to Control Model Expressivity and Improve Optimization
- Symmetries, flat minima, and the conserved quantities of gradient flow
- Explaining Near-Zero Hessian Eigenvalues Through Approximate Symmetries in Neural Networks
Related papers
- Topological metal-insulator transitions in one-dimensional non-Hermitian quasicrystals: beyond PT-symmetry
- Microscopic and collective signatures of feature learning in neural networks
- Synchronous and asynchronous Monte Carlo dynamics of mean-field spin glasses with rotationally invariant couplings: what they share and where they differ
- Reentrant Localization Length in a Quasiperiodic Thue-Morse Chain
- Beyond Ground States: Learning Many-Body Hamiltonians for p-Bit Invertible Arithmetic
- Linear equivalence of nonlinear recurrent neural networks