Guest Lecture by Peter Lory: Acceleration of Two-Party Protocols Using Chebyshev Polynomials
Nov 12, 1pm @SBA
The lecture focuses on the potential of Chebyshev polynomials for privacy-enhancing techniques in data mining. Two protocols are considered: 1) The calculation of the arithmetic mean according to Kiltz, Leander and Malone-Lee (2005). 2) The secure distributed learning of decision trees according to Lindell and Pinkas (2002). The studies show that significant efficiency gains can be achieved in both protocols through the use of Chebyshev polynomials. This is due to the better approximation properties of aborted Chebyshev developments compared to the aborted Taylor series.