This position paper argues that parameter symmetry breaking and restoration unify three hierarchies in deep learning: learning dynamics, model complexity, and representation formation.
Neural thermodynamics i: Entropic forces in deep and universal representation learning
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
support 1representative citing papers
citing papers explorer
-
Parameter Symmetry Potentially Unifies Deep Learning Theory
This position paper argues that parameter symmetry breaking and restoration unify three hierarchies in deep learning: learning dynamics, model complexity, and representation formation.