Algorithms for solving integral loss minimization problems. Currently, we handle the following problems:
- Joint Quantile Regression
- Emotion Transfer with squared loss
- ITL with general loss provided by the user based on pytorch autodiff
To install the package, clone it and run
$ pip install -e .
This installs automatically the non-satisfied dependencies in ./requirements.txt
You can find examples for both problems in the demo section, for both infinite quantile regression and emotion transfer (with a bonus video).
If you use this code, please cite the corresponding work:
@inproceedings{brault2019infinite,
title={Infinite task learning in rkhss},
author={Brault, Romain and Lambert, Alex and Szab{\'o}, Zolt{\'a}n and Sangnier, Maxime and d’Alch{\'e}-Buc, Florence},
booktitle={The 22nd International Conference on Artificial Intelligence and Statistics},
pages={1294--1302},
year={2019},
organization={PMLR}
}
@preprint{emo_transfer_lambert,
title={Emotion Transfer Using Vector-Valued Infinite Task Learning},
author={Lambert, Alex and Parekh, Sanjeel and Szab{\'o}, Zolt{\'a}n and d’Alch{\'e}-Buc, Florence},
year={2021}
}