TY - JOUR
T1 - Metric-Fair Active Learning
AU - Shen, Jie
AU - Cui, Nan
AU - Wang, Jing
N1 - Publisher Copyright:
Copyright © 2022 by the author(s)
PY - 2022
Y1 - 2022
N2 - Active learning has become a prevalent technique for designing label-efficient algorithms, where the central principle is to only query and fit “informative” labeled instances. It is, however, known that an active learning algorithm may incur unfairness due to such instance selection procedure. In this paper, we henceforth study metric-fair active learning of homogeneous halfspaces, and show that under the distribution-dependent PAC learning model, fairness and label efficiency can be achieved simultaneously. We further propose two extensions of our main results: 1) we show that it is possible to make the algorithm robust to the adversarial noise - one of the most challenging noise models in learning theory; and 2) it is possible to significantly improve the label complexity when the underlying halfspace is sparse.
AB - Active learning has become a prevalent technique for designing label-efficient algorithms, where the central principle is to only query and fit “informative” labeled instances. It is, however, known that an active learning algorithm may incur unfairness due to such instance selection procedure. In this paper, we henceforth study metric-fair active learning of homogeneous halfspaces, and show that under the distribution-dependent PAC learning model, fairness and label efficiency can be achieved simultaneously. We further propose two extensions of our main results: 1) we show that it is possible to make the algorithm robust to the adversarial noise - one of the most challenging noise models in learning theory; and 2) it is possible to significantly improve the label complexity when the underlying halfspace is sparse.
UR - http://www.scopus.com/inward/record.url?scp=85163067816&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85163067816&partnerID=8YFLogxK
M3 - Conference article
AN - SCOPUS:85163067816
VL - 162
SP - 19809
EP - 19826
JO - Proceedings of Machine Learning Research
JF - Proceedings of Machine Learning Research
T2 - 39th International Conference on Machine Learning, ICML 2022
Y2 - 17 July 2022 through 23 July 2022
ER -