TY - JOUR
T1 - Robust Learning from Discriminative Feature Feedback
AU - Dasgupta, Sanjoy
AU - Sabato, Sivan
N1 - Funding Information:
This research was supported by National Science Foundation grant CCF-1813160, and by a United-States-Israel Binational Science Foundation (BSF) grant no. 2017641. Part of the work was done while the authors were at the “Foundations of Machine Learning” program at the Simons Institute for the Theory of Computing, Berkeley.
Publisher Copyright:
Copyright © 2020 by the author(s)
PY - 2020/1/1
Y1 - 2020/1/1
N2 - Recent work introduced the model of learning from discriminative feature feedback, in which a human annotator not only provides labels of instances, but also identifies discriminative features that highlight important differences between pairs of instances. It was shown that such feedback can be conducive to learning, and makes it possible to efficiently learn some concept classes that would otherwise be intractable. However, these results all relied upon perfect annotator feedback. In this paper, we introduce a more realistic, robust version of the framework, in which the annotator is allowed to make mistakes. We show how such errors can be handled algorithmically, in both an adversarial and a stochastic setting. In particular, we derive regret bounds in both settings that, as in the case of a perfect annotator, are independent of the number of features. We show that this result cannot be obtained by a naive reduction from the robust setting to the non-robust setting.
AB - Recent work introduced the model of learning from discriminative feature feedback, in which a human annotator not only provides labels of instances, but also identifies discriminative features that highlight important differences between pairs of instances. It was shown that such feedback can be conducive to learning, and makes it possible to efficiently learn some concept classes that would otherwise be intractable. However, these results all relied upon perfect annotator feedback. In this paper, we introduce a more realistic, robust version of the framework, in which the annotator is allowed to make mistakes. We show how such errors can be handled algorithmically, in both an adversarial and a stochastic setting. In particular, we derive regret bounds in both settings that, as in the case of a perfect annotator, are independent of the number of features. We show that this result cannot be obtained by a naive reduction from the robust setting to the non-robust setting.
UR - http://www.scopus.com/inward/record.url?scp=85161911295&partnerID=8YFLogxK
M3 - Conference article
AN - SCOPUS:85161911295
SN - 2640-3498
VL - 108
SP - 973
EP - 982
JO - Proceedings of Machine Learning Research
JF - Proceedings of Machine Learning Research
T2 - 23rd International Conference on Artificial Intelligence and Statistics, AISTATS 2020
Y2 - 26 August 2020 through 28 August 2020
ER -