Strategic Classification with Unknown User Manipulations

Tosca Lechner, Ruth Urner, Shai Ben-David
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:18714-18732, 2023.

Abstract

In many human-centric applications for Machine Learning instances will adapt to a classifier after its deployment. The field of strategic classification deals with this issue by aiming for a classifier that balances the trade-off between correctness and robustness to manipulation. This task is made harder if the underlying manipulation structure (i.e. the set of manipulations available at every instance) is unknown to the learner. We propose a novel batch-learning setting in which we use unlabeled data from previous rounds to estimate the manipulation structure. We show that in this batch-learning setting it is possible to learn a close-to-optimal classifier in terms of the strategic loss even without knowing the feasible manipulations beforehand. In line with recent advances in the strategic classification literature, we do not assume a best-response from agents but only require that observed manipulations are feasible.

Cite this Paper

BibTeX
@InProceedings{pmlr-v202-lechner23a, title = {Strategic Classification with Unknown User Manipulations}, author = {Lechner, Tosca and Urner, Ruth and Ben-David, Shai}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {18714--18732}, year = {2023}, editor = {Krause, Andreas and Brunskill, Emma and Cho, Kyunghyun and Engelhardt, Barbara and Sabato, Sivan and Scarlett, Jonathan}, volume = {202}, series = {Proceedings of Machine Learning Research}, month = {23--29 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v202/lechner23a/lechner23a.pdf}, url = {https://proceedings.mlr.press/v202/lechner23a.html}, abstract = {In many human-centric applications for Machine Learning instances will adapt to a classifier after its deployment. The field of strategic classification deals with this issue by aiming for a classifier that balances the trade-off between correctness and robustness to manipulation. This task is made harder if the underlying manipulation structure (i.e. the set of manipulations available at every instance) is unknown to the learner. We propose a novel batch-learning setting in which we use unlabeled data from previous rounds to estimate the manipulation structure. We show that in this batch-learning setting it is possible to learn a close-to-optimal classifier in terms of the strategic loss even without knowing the feasible manipulations beforehand. In line with recent advances in the strategic classification literature, we do not assume a best-response from agents but only require that observed manipulations are feasible.} }
Endnote
%0 Conference Paper %T Strategic Classification with Unknown User Manipulations %A Tosca Lechner %A Ruth Urner %A Shai Ben-David %B Proceedings of the 40th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2023 %E Andreas Krause %E Emma Brunskill %E Kyunghyun Cho %E Barbara Engelhardt %E Sivan Sabato %E Jonathan Scarlett %F pmlr-v202-lechner23a %I PMLR %P 18714--18732 %U https://proceedings.mlr.press/v202/lechner23a.html %V 202 %X In many human-centric applications for Machine Learning instances will adapt to a classifier after its deployment. The field of strategic classification deals with this issue by aiming for a classifier that balances the trade-off between correctness and robustness to manipulation. This task is made harder if the underlying manipulation structure (i.e. the set of manipulations available at every instance) is unknown to the learner. We propose a novel batch-learning setting in which we use unlabeled data from previous rounds to estimate the manipulation structure. We show that in this batch-learning setting it is possible to learn a close-to-optimal classifier in terms of the strategic loss even without knowing the feasible manipulations beforehand. In line with recent advances in the strategic classification literature, we do not assume a best-response from agents but only require that observed manipulations are feasible.
APA
Lechner, T., Urner, R. & Ben-David, S.. (2023). Strategic Classification with Unknown User Manipulations. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:18714-18732 Available from https://proceedings.mlr.press/v202/lechner23a.html.

Related Material