AI Paper Atlas

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

More PAC-Bayes bounds: From bounded losses, to losses with general tail behaviors, to anytime validity

Borja Rodríguez-Gálvez, Ragnar Thobaben, Mikael Skoglund

JMLR 2024 · Learning theory and algorithms › Kernels and regression theory · approach: PAC-Bayes bounds · 9 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 the proposed fast/mixed-rate bound objective instead of the classical bound objectivePAC-Bayes with backprop · MNIST, CNN, PAC-Bayes with backproppopulation risk certificate: 0.23658 -> 0.17501Improved
use the proposed bound objective instead of previous bound objective 1PAC-Bayes with backprop · MNIST, CNN, PAC-Bayes with backproppopulation risk certificate: 0.20870 -> 0.17501Improved
use the proposed bound objective instead of previous bound objective 2PAC-Bayes with backprop · MNIST, CNN, PAC-Bayes with backproppopulation risk certificate: 0.21159 -> 0.17501Improved
use the second proposed objective instead of previous bound objective 1PAC-Bayes with backprop · MNIST, CNN, PAC-Bayes with backproppopulation risk certificate: 0.20870 -> 0.19763Improved
use the proposed objective instead of the classical objectivePAC-Bayes with backprop · MNIST, CNN, PAC-Bayes with backpropempirical risk: 0.23658 -> 0.07054Improved
use the proposed objective instead of the classical objectivePAC-Bayes with backprop · MNIST, CNN, PAC-Bayes with backpropnormalized dependency D/n: 0.02715 -> 0.04649Hurt
use hard-coded Seeger-Langford gradients instead of the proposed fast-rate objectivePAC-Bayes with backprop · MNIST, CNN, PAC-Bayes with backproppopulation risk certificate: 0.17501 -> 0.16922Improved

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