Skip to content

gauthierthurin/OTCP

Repository files navigation

This repo contains codes for our paper on Optimal-Transport based conformal prediction, available at https://arxiv.org/abs/2501.18991v2

functions.py contains our main functions

Classification

Classification.ipynb contains experiments for Classification in the main body of the paper.

Classification2.ipynb contains experiments for Classification in the supplementary material for K=5 labels.

Regression

The file 'data' contains the real datasets that were downloaded from https://github.com/tsoumakas/mulan/blob/master/data/multi-target/README.md

Regression.ipynb contains experiments for Regression, on simulated data and on real data.

Files taken from previous works

Experiments in classification require the package arc for the ARS method, available at: https://github.com/msesia/arc Folder : arc

Experiments in regression involve the local ellipsoid method, available at: https://github.com/M-Soundouss/EllipsoidalConformalMTR/tree/main File : ellipsoidal_conformal_utilities.py

About

Optimal-transport based conformal prediction

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors