Adversarial Multi Class Learning under Weak Supervision with Performance Guarantees

Alessio Mazzetto, Cyrus Cousins, Dylan Sam, Stephen H Bach, Eli Upfal
Proceedings of the 38th International Conference on Machine Learning, PMLR 139:7534-7543, 2021.

Abstract

We develop a rigorous approach for using a set of arbitrarily correlated weak supervision sources in order to solve a multiclass classification task when only a very small set of labeled data is available. Our learning algorithm provably converges to a model that has minimum empirical risk with respect to an adversarial choice over feasible labelings for a set of unlabeled data, where the feasibility of a labeling is computed through constraints defined by rigorously estimated statistics of the weak supervision sources. We show theoretical guarantees for this approach that depend on the information provided by the weak supervision sources. Notably, this method does not require the weak supervision sources to have the same labeling space as the multiclass classification task. We demonstrate the effectiveness of our approach with experiments on various image classification tasks.

Cite this Paper

BibTeX
@InProceedings{pmlr-v139-mazzetto21a, title = {Adversarial Multi Class Learning under Weak Supervision with Performance Guarantees}, author = {Mazzetto, Alessio and Cousins, Cyrus and Sam, Dylan and Bach, Stephen H and Upfal, Eli}, booktitle = {Proceedings of the 38th International Conference on Machine Learning}, pages = {7534--7543}, year = {2021}, editor = {Meila, Marina and Zhang, Tong}, volume = {139}, series = {Proceedings of Machine Learning Research}, month = {18--24 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v139/mazzetto21a/mazzetto21a.pdf}, url = {https://proceedings.mlr.press/v139/mazzetto21a.html}, abstract = {We develop a rigorous approach for using a set of arbitrarily correlated weak supervision sources in order to solve a multiclass classification task when only a very small set of labeled data is available. Our learning algorithm provably converges to a model that has minimum empirical risk with respect to an adversarial choice over feasible labelings for a set of unlabeled data, where the feasibility of a labeling is computed through constraints defined by rigorously estimated statistics of the weak supervision sources. We show theoretical guarantees for this approach that depend on the information provided by the weak supervision sources. Notably, this method does not require the weak supervision sources to have the same labeling space as the multiclass classification task. We demonstrate the effectiveness of our approach with experiments on various image classification tasks.} }
Endnote
%0 Conference Paper %T Adversarial Multi Class Learning under Weak Supervision with Performance Guarantees %A Alessio Mazzetto %A Cyrus Cousins %A Dylan Sam %A Stephen H Bach %A Eli Upfal %B Proceedings of the 38th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2021 %E Marina Meila %E Tong Zhang %F pmlr-v139-mazzetto21a %I PMLR %P 7534--7543 %U https://proceedings.mlr.press/v139/mazzetto21a.html %V 139 %X We develop a rigorous approach for using a set of arbitrarily correlated weak supervision sources in order to solve a multiclass classification task when only a very small set of labeled data is available. Our learning algorithm provably converges to a model that has minimum empirical risk with respect to an adversarial choice over feasible labelings for a set of unlabeled data, where the feasibility of a labeling is computed through constraints defined by rigorously estimated statistics of the weak supervision sources. We show theoretical guarantees for this approach that depend on the information provided by the weak supervision sources. Notably, this method does not require the weak supervision sources to have the same labeling space as the multiclass classification task. We demonstrate the effectiveness of our approach with experiments on various image classification tasks.
APA
Mazzetto, A., Cousins, C., Sam, D., Bach, S.H. & Upfal, E.. (2021). Adversarial Multi Class Learning under Weak Supervision with Performance Guarantees. Proceedings of the 38th International Conference on Machine Learning, in Proceedings of Machine Learning Research 139:7534-7543 Available from https://proceedings.mlr.press/v139/mazzetto21a.html.

Related Material