站内直航

学术报告：From Mental Poker to Core Business: Why and How to Deploy Secure Computation Protocols?

Title: From Mental Poker to Core Business: Why and How to Deploy Secure Computation Protocols?

时间：2015年11月5日  上午9:30-10:30

地点：中国科学院信息工程研究所3号楼3224会议室。

报告人：Prof. Moti Yung(Google Inc. and Columbia University)

Abstract:

Technological innovations in security and privacy are critical to advancing modern computing in our time. I will present an effort involving deployment of experimental commercial applications designed and built as a 'secure multi-party computation protocol for specific tasks,' to be used repetitively to achieve a number of concrete ubiquitous business goals. In these applications, the outputs are calculated in the presence of privacy constraints which prevent parties from sharing their individual inputs directly and openly. I will also discuss what I think are the reasons for the inherent difficulty of developing such routines in general (for achieving business goals). In particular, I will survey what I believe to be the reasons that almost 40 years since secure computation protocols was invented as a basic theoretical notion, capturing specific and then general computational tasks, and in spite of its theoretical and even experimentation success, the notion has not yet been widely and seriously used in achieving routine relevant business goals (in contrast with symmetric key and public key cryptosystems and protocols, which were also proposed 40 years ago and are used extensively, primarily to implement secure authenticated channels). The presentation will also cover the general bottom up methodology used in this effort leading to the design and development process. This exemplifying methodology includes: feasibility study of the specific domain, extraction of business needs which are limited by privacy constraints, application analysis from the perspective of utility metrics and secure computing. Then, the methodology further includes design, implementation, and experimentation, guided by the analysis and employing appropriate protocols, while considering scale and performance constraints, and cost overhead that is tolerable.