CORTEXA
← Browse
arxivcond-mat.dis-nncond-mat.stat-mechcs.LG2026-07-05

Broken Ergodicity and the Violation of the Fluctuation-Dissipation Theorem Lead to Generalization Beyond Overfitting in Machine Learning

Chan Li, Nigel Goldenfeld

The remarkable ability of modern neural networks to generalize improves with increasing network capacity, even when the number of model parameters or effective degrees of freedom exceeds the number of training data points. This phenomenon is all the more surprising given that generalization error diverges when the number of model parameters approaches a critical value from below. Here we use dynamical mean field theory to show that this so-called "double descent" behavior is the outcome of a phase transition in the stochastic field theory describing the training process. We calculate the critical exponents and scaling function of the double descent phase transition, and show that it is marked by a breakdown of the fluctuation-dissipation theorem associated with broken ergodicity. The corresponding response function has the same functional form as the simple London model of the superconducting transition, with the rigidity of the wave function corresponding to the neural network's ability to generalize accurately.

View free PDFSource page

Related papers

arxivcond-mat.dis-nncond-mat.stat-mechcs.AInlin.CD2026-07-24

Multiplicity of Stable Attractors in Disordered Neural Models

Raffaele Marino, Roberto Livi, Antonio Politi

We show how large-deviation statistics allows one to obtain reliable estimates of the multiplicity of stable fixed-points in a model of neural ordinary differential equations previously employed in computational tasks. The result is obtained by developing a suitable perturbative…

View free PDFSource page
arxivphysics.flu-dyncs.LGquant-ph2026-07-23

Explainable quantum-compressed machine learning for complex fluid flows

Xiao Xue, Maida Wang, Mingyang Gao, Minh Chung, Peter V. Coveney

Machine-learning surrogates of physical systems face a paradox: explainable models facing the challenge of expressivity to capture complex nonlinear flows, whereas expressive deep surrogates match high-fidelity simulations only through massive parameterisations that turn the lear…

View free PDFSource page
arxivcs.LGastro-ph.COastro-ph.GAhep-exhep-phstat.ML2026-07-23

An Introduction to Bayesian and Frequentist Simulation-Based Inference with Machine Learning

Maximilian Dax, Theo Heimel, Gilles Louppe

Simulation-based inference (SBI) with machine learning is an increasingly important tool for solving inverse problems in science and engineering, including parameter inference and the inversion of detector effects. We provide an overview of the Bayesian and frequentist statistica…

View free PDFSource page
arxivstat.MLcs.LGstat.AP2026-07-31

Analytical and Bootstrap Confidence Intervals of Double Machine Learning: Simulation studies and an application to rural-urban difference in obesity prevalence

Haozheng Xu, Siyuan Ma, Qingyan Xiang

Double Machine Learning (DML) is a popular approach for treatment effect estimation in various settings, which allows a wide range of flexible machine learning methods to be used for nuisance parameter estimation while preserving valid inference. In practice, however, applied res…

View free PDFSource page
arxivcs.LGcs.AI2026-07-31

Implicit Machine Learning Force Fields Accelerate Molecular Dynamics Simulations

Johannes Maeß, Leon Werner, J. Thorben Frank, Winfried Ripken, Martin Michajlow, Joshua Futterer, et al.

We introduce implicit machine learning force fields (I-MLFFs), which replace explicit stacks of neural network layers with self-consistent fixed-point equations. In molecular simulations, this formulation enables intermediate representations to be reused across successive timeste…

View free PDFSource page