README.md

Exact Optimization of Conformal Predictors via Incremental and Decremental Learning

Implementation and evaluation of full CP (classifiers and regressors) optimized via incremental&decremental learning.

Paper at: https://arxiv.org/abs/2102.03236. To appear: ICML '21.

Code structure

Please, refer to the notebook exact-cp-optimization.ipynb for implementation and evaluation details. If you have troubles viewing the notebook via Github, use: https://nbviewer.jupyter.org/github/gchers/exact-cp-optimization/blob/main/exact-cp-optimization.ipynb.

The code in eli.py is a simple library used for carrying out the experiments.

Citing

@misc{cherubin2021exact,
      title={Exact Optimization of Conformal Predictors via Incremental and Decremental Learning}, 
      author={Giovanni Cherubin and Konstantinos Chatzikokolakis and Martin Jaggi},
      year={2021},
      eprint={2102.03236},
      archivePrefix={arXiv},
      primaryClass={cs.LG}
}