AI Paper Atlas

Research map / Learning theory and algorithms / Sample complexity

CS4ML: A general framework for active learning with arbitrary data based on Christoffel functions

Juan M Cardenas, Ben Adcock, Nick Dexter

NeurIPS 2023 · Learning theory and algorithms › Sample complexity · approach: PAC learnability theory · 1 citations (OpenAlex)

Read the paper

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
sample with Christoffel-function-based measure (CS) instead of Monte Carlo samplinggradient-augmented polynomial regression · exp(-(theta_1+...+theta_d)/(2d)), d=2,4,8, log-linear maverage regression errorImproved
sample frequency lines with CS instead of uniform random samplinggenerative-model MRI reconstruction · 3D brain MRI, 0.125%, 0.375%, 0.625% samplingPSNRImproved
use Christoffel Adaptive Sampling (CAS) instead of Monte Carlo samplingPINN for Burgers equation · tanh and sigmoid 5x50 DNNstest error vs number of samplesImproved
use empirical approximation of the optimal sampling measure instead of exact samplingCS for generative models · MRI reconstructionPSNRImproved
apply CS in cases outside the theory (nonlinear PDE operator, adaptive scheme)CAS PINNs · Burgers equationtest errorImproved
use log-linear scaling of m with n for MCSMCS gradient-augmented polynomial regression · hyperbolic cross polynomial spacesgeneralization errorHurt
apply CS in other settingsCS4ML · appendix caseserrorNo clear effect

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