Asymptotics of representation learning in finite Bayesian neural networks*

DOI10.1088/1742-5468/AC98A6MaRDI QIDQ5055422zbMATH OpenOpenAlexFDO

Authors Jacob A. Zavatone-Veth, Abdulkadir Canatar, Benjamin S. Ruben, Cengiz Pehlevan

Publication date 13 December 2022

Published in Journal of Statistical Mechanics: Theory and Experiment (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/2106.00651

zbMATH Keywords

machine learning

Mathematics Subject Classification ID

Statistical mechanics, structure of matter (82-XX)

Recommendations

• Wide neural networks of any depth evolve as linear models under gradient descent ^*
• Unified field theoretical approach to deep and recurrent neuronal networks
• Learning sparse features can lead to overfitting in neural networks
• Infinite-width limit of deep linear neural networks

Cites work

• scientific article; zbMATH DE number 54145 (Why is no real title available?)
• Bayesian learning for neural networks
• Deep learning
• The Evaluation of the Collision Matrix
• The Principles of Deep Learning Theory

Cited in

(5)

• Bayesian Representation of Stochastic Processes under Learning: de Finetti Revisited
• Learning curves for deep structured Gaussian feature models
• Dynamics of finite width kernel and prediction fluctuations in mean field neural networks
• Bayes-optimal learning of deep random networks of extensive-width
• A primer on Bayesian neural networks: review and debates

Describes a project that uses

Uses Software

• PyTorch

This page was built for publication: Asymptotics of representation learning in finite Bayesian neural networks*

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5055422)