Research map / Learning theory and algorithms / Kernels and regression theory
Bayesian Treatment of the Spectrum of the Empirical Kernel in (Sub)Linear-Width Neural Networks
Ouns El Harzli, Bernardo Cuenca Grau
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Theoretical study of Bayesian neural networks in the joint limit of growing training set, width and input dimension, linking Mercer eigenvalues of the empirical kernel to limiting spectral distributions from random matrix theory.
Problem
Understanding Bayesian neural networks in the theoretical limit where the number of training examples, network width, and input dimension all grow, including sub-linear width regimes.
Approach
Establishes a correspondence between Mercer eigenvalues of the empirical kernel and limiting spectral distributions of covariance matrices from random matrix theory, bridging kernel-theoretic approaches and statistical mechanics for BNNs.
Closest papers by meaning
- Critical feature learning in deep neural networks (ICML 2024 · similarity 0.89)
- A Study of Bayesian Neural Network Surrogates for Bayesian Optimization (ICLR 2024 · similarity 0.88)
- When narrower is better: the narrow width limit of Bayesian parallel branching neural networks (ICLR 2025 · similarity 0.88)
- Training Bayesian Neural Networks with Sparse Subspace Variational Inference (ICLR 2024 · similarity 0.88)
- Bayes-optimal learning of an extensive-width neural network from quadratically many samples (NeurIPS 2024 · similarity 0.88)
- An Infinite-Width Analysis on the Jacobian-Regularised Training of a Neural Network (ICML 2024 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
