Theory of Computing
-------------------
Title : Private Learning and Sanitization: Pure vs. Approximate Differential Privacy
Authors : Amos Beimel, Kobbi Nissim, and Uri Stemmer
Volume : 12
Number : 1
Pages : 1-61
URL : http://www.theoryofcomputing.org/articles/v012a001
Abstract
--------
We compare the sample complexity of private learning [Kasiviswanathan
et al. 2008] and sanitization [Blum et al. 2008] under _pure_
$\epsilon$-differential privacy [Dwork et al. TCC 2006] and
_approximate_ $(\epsilon,\delta)$-differential privacy [Dwork et al.
Eurocrypt 2006]. We show that the sample complexity of these tasks
under approximate differential privacy can be significantly lower than
that under pure differential privacy.
We define a family of optimization problems, which we call _Quasi-
Concave Promise Problems_, that generalizes some of the tasks we
consider. We observe that a quasi-concave promise problem can be
privately approximated using a solution to a smaller instance of a
quasi-concave promise problem. This allows us to construct an
efficient recursive algorithm to solve such problems privately.
Specifically, we construct private learners for point functions,
threshold functions, and axis-aligned rectangles in high dimension.
Similarly, we construct sanitizers for point functions and threshold
functions.
We also examine the sample complexity of _label-private_ learners, a
relaxation of private learning where the learner is required to only
protect the privacy of the labels in the sample. We show that the VC
dimension completely characterizes the sample complexity of such
learners, that is, the sample complexity of learning with label
privacy is equal (up to constants) to learning without privacy.
An extended abstract of this paper appeared in the Proceedings of the
17th International Workshop on Randomization and Computation, 2013.