2023-08-17-rodriguez23a.md

title	abstract	layout	series	publisher	issn	id	month	tex_title	firstpage	lastpage	page	order	cycles	bibtex_author	author	date	address	container-title	volume	genre	issued	pdf	extras
Self Learning using Venn-Abers predictors	In supervised learning problems, it is common to have a lot of unlabeled data, but little labeled data. It is then desirable to leverage the unlabeled data to improve the learning procedure. One way to do this is to have a model predict “pseudolabels” for the unlabeled data, so as to use them for learning. In self-learning, the pseudo-labels are provided by the very same model to which they are fed. As these pseudo-labels are by nature uncertain and only partially reliable, it is then natural to model this uncertainty and take it into account in the learning process, if only to robustify the self-learning procedure. This paper describes such an approach, where we use Venn-Abers Predictors to produce calibrated credal labels so as to quantify the pseudo-labeling uncertainty. These labels are then included in the learning process by optimizing an adapted loss. Experiments show that taking into account pseudo-label uncertainty both robustifies the self-learning procedure and allows it to converge faster in general.	inproceedings	Proceedings of Machine Learning Research	PMLR	2640-3498	rodriguez23a	0	Self Learning using Venn-Abers predictors	234	250	234-250	234	false	Rodriguez, Come and Martin Bordini, Vitor and Destercke, Sebastien and Quost, Benjamin	given family Come Rodriguez given family Vitor Martin Bordini given family Sebastien Destercke given family Benjamin Quost	2023-08-17		Proceedings of the Twelfth Symposium on Conformal and Probabilistic Prediction with Applications	204	inproceedings	date-parts 2023 8 17	https://proceedings.mlr.press/v204/rodriguez23a/rodriguez23a.pdf