AI Paper Atlas

Research map / Learning theory and algorithms / Kernels and regression theory

Generalization of Gibbs and Langevin Monte Carlo Algorithms in the Interpolation Regime

Andreas Maurer, Erfan Mirzaei, Massimiliano Pontil

ICML 2026 · Learning theory and algorithms › Kernels and regression theory · approach: Information-theoretic generalization bounds · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use calibrated Gibbs-posterior bound instead of self-certified f_quad certificateFCN on MNIST · MNIST 10-class FCNtrue 0-1 risk bound: 0.3155 -> 0.2470Improved
use calibrated Gibbs-posterior bound instead of f_quad certificateCNN on MNIST · MNIST 10-class CNNtrue 0-1 risk bound: 0.2165 -> 0.2025Improved
use SGLD-trained standard model instead of f_bbbFCN on MNIST · MNIST 10-class FCNtrue 0-1 risk bound: 0.5516 -> 0.2470Improved
use SGLD-trained standard model instead of f_bbbFCN on MNIST · MNIST 10-class FCNtest 0-1 error: 0.0293 -> 0.0338Hurt
use SGLD-trained standard model instead of f_bbbCNN on MNIST · MNIST 10-class CNNtest 0-1 error: 0.0154 -> 0.0182Hurt
lower temperature (beta 500 to 1000) in SGLD with true labelsSGLD fully connected net · MNIST binarytrain error: 0.17 -> 0.1Improved
use random labels instead of true labels at beta=64000SGLD · MNIST binarytrain error and bound: train error about 0.1 while bound above 0.5Mixed

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy