Truthful Linear Regression

Rachel Cummings, Stratis Ioannidis, Katrina Ligett
Proceedings of The 28th Conference on Learning Theory, PMLR 40:448-483, 2015.

Abstract

We consider the problem of fitting a linear model to data held by individuals who are concerned about their privacy. Incentivizing most players to truthfully report their data to the analyst constrains our design to mechanisms that provide a privacy guarantee to the participants; we use differential privacy to model individuals’ privacy losses. This immediately poses a problem, as differentially private computation of a linear model necessarily produces a biased estimation, and existing approaches to design mechanisms to elicit data from privacy-sensitive individuals do not generalize well to biased estimators. We overcome this challenge through an appropriate design of the computation and payment scheme.

Cite this Paper

BibTeX
@InProceedings{pmlr-v40-Cummings15, title = {Truthful Linear Regression}, author = {Cummings, Rachel and Ioannidis, Stratis and Ligett, Katrina}, booktitle = {Proceedings of The 28th Conference on Learning Theory}, pages = {448--483}, year = {2015}, editor = {Grünwald, Peter and Hazan, Elad and Kale, Satyen}, volume = {40}, series = {Proceedings of Machine Learning Research}, address = {Paris, France}, month = {03--06 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v40/Cummings15.pdf}, url = {https://proceedings.mlr.press/v40/Cummings15.html}, abstract = {We consider the problem of fitting a linear model to data held by individuals who are concerned about their privacy. Incentivizing most players to truthfully report their data to the analyst constrains our design to mechanisms that provide a privacy guarantee to the participants; we use differential privacy to model individuals’ privacy losses. This immediately poses a problem, as differentially private computation of a linear model necessarily produces a biased estimation, and existing approaches to design mechanisms to elicit data from privacy-sensitive individuals do not generalize well to biased estimators. We overcome this challenge through an appropriate design of the computation and payment scheme.} }
Endnote
%0 Conference Paper %T Truthful Linear Regression %A Rachel Cummings %A Stratis Ioannidis %A Katrina Ligett %B Proceedings of The 28th Conference on Learning Theory %C Proceedings of Machine Learning Research %D 2015 %E Peter Grünwald %E Elad Hazan %E Satyen Kale %F pmlr-v40-Cummings15 %I PMLR %P 448--483 %U https://proceedings.mlr.press/v40/Cummings15.html %V 40 %X We consider the problem of fitting a linear model to data held by individuals who are concerned about their privacy. Incentivizing most players to truthfully report their data to the analyst constrains our design to mechanisms that provide a privacy guarantee to the participants; we use differential privacy to model individuals’ privacy losses. This immediately poses a problem, as differentially private computation of a linear model necessarily produces a biased estimation, and existing approaches to design mechanisms to elicit data from privacy-sensitive individuals do not generalize well to biased estimators. We overcome this challenge through an appropriate design of the computation and payment scheme.
RIS
TY - CPAPER TI - Truthful Linear Regression AU - Rachel Cummings AU - Stratis Ioannidis AU - Katrina Ligett BT - Proceedings of The 28th Conference on Learning Theory DA - 2015/06/26 ED - Peter Grünwald ED - Elad Hazan ED - Satyen Kale ID - pmlr-v40-Cummings15 PB - PMLR DP - Proceedings of Machine Learning Research VL - 40 SP - 448 EP - 483 L1 - http://proceedings.mlr.press/v40/Cummings15.pdf UR - https://proceedings.mlr.press/v40/Cummings15.html AB - We consider the problem of fitting a linear model to data held by individuals who are concerned about their privacy. Incentivizing most players to truthfully report their data to the analyst constrains our design to mechanisms that provide a privacy guarantee to the participants; we use differential privacy to model individuals’ privacy losses. This immediately poses a problem, as differentially private computation of a linear model necessarily produces a biased estimation, and existing approaches to design mechanisms to elicit data from privacy-sensitive individuals do not generalize well to biased estimators. We overcome this challenge through an appropriate design of the computation and payment scheme. ER -
APA
Cummings, R., Ioannidis, S. & Ligett, K.. (2015). Truthful Linear Regression. Proceedings of The 28th Conference on Learning Theory, in Proceedings of Machine Learning Research 40:448-483 Available from https://proceedings.mlr.press/v40/Cummings15.html.

Related Material