an:06168649
Zbl 1270.90029
Fung, Glenn M.; Mangasarian, Olvi L.
Privacy-preserving linear and nonlinear approximation via linear programming
EN
Optim. Methods Softw. 28, No. 1, 207-216 (2013).
00317871
2013
j
90C05 90C90
privacy-preserving approximation; random kernels; support vector machines; linear programming
Summary: We propose a novel privacy-preserving random kernel approximation based on a data matrix \(A\in \mathbb{R}^{m\times n}\) whose rows are divided into privately owned blocks. Each block of rows belongs to a different entity that is unwilling to share its rows or make them public. We wish to obtain an accurate function approximation for a given \(y\in \mathbb{R}^m\) corresponding to each of the \(m\) rows of \(A\). Our approximation of \(y\) is a real function on \(\mathbb{R}^n\) evaluated at each row of \(4\) and is based on the concept of a reduced kernel \(K(A, B^{\prime})\), where \(B^\prime\) is the transpose of a completely random matrix \(B\). The proposed linear-programming-based approximation, which is public but does not reveal the privately held data matrix \(A\), has an accuracy comparable to that of an ordinary kernel approximation based on a publicly disclosed data matrix \(A\).