Secure data processing: perspectives and challenges

Prof. Renaud Sirdey

Renaud Sirdey

CEA LIST

Saclay, France




Abstract:

Secure data communication and storage has been studied for a long time and it is fair to say that this problem is now well understood and, to some extent, solved. However, with the blurring of the physical relationship between the user and the physical hardware that will eventually process its data, there is a growing need to provide not only private communication and secure storage but to further protect user data while they are being processed. In this talk, we will cover techniques for secure data processing ranging from pragmatic hardware-assisted solutions to promising software-only techniques based on currently emerging homomorphic encryption primitives.

Short Bio:

R. Sirdey, PhD, HDR, is a senior researcher at Commissariat à l'Energie Atomique, the French DoE, where he leads the Embedded Real-Time Systems Laboratory. His main research interests include parallelism, compilation, discrete optimization as well as applied cryptology. Prior to his current responsibilities, after spending around 10 years as a system architect in the telecom industry, he has most notably led the research team which designed a complete industry-grade dataflow compiler for the 256 cores MPPA architecture as part of a joint CEA/KALRAY lab. More recently, while leading a CEA research project aiming at prototyping a hardware-assisted security-hardened IaaS solution, he has also started to work on compilation and parallel execution supports for homomorphic cryptocalculations.

Optimization problems for efficient exploitation of high-performance computing systems

Julius Zilinskas

Vilnius University

Lithuania




Abstract:

Current (and likely future) high-performance computing systems are hierarchical and often heterogeneous. Therefore communication and computing speed as well as energy consumption may be non-uniform. Because of this, performance of parallel applications depends on location of processes within the system. Efficient exploitation of high-performance computing systems enables considerable saving of resources what is nowadays increasingly important. In this talk we formulate and discuss some related optimization problems and ways to solve them.

Short Bio:

J. Zilinskas is a principal researcher and a head of department at Vilnius University, Lithuania. He received PhD at Kaunas University of Technology, Lithuania in 2002. After Royal Society/NATO postdoctoral fellowship at University College London in 2003-2004, he received NATO reintegration grant to continue research career in Lithuania at the Institute of Mathematics and Informatics, which became part of Vilnius University in 2010. Research interests of J. Zilinskas include parallel computing, optimization, and multidimensional data visualization. He is a nominated expert of Lithuania at COST domain Information and Communication Technologies and a member of various bodies: European Science Foundation (ESF) Standing Committee for Physical and Engineering Sciences (PESC), management committee of COST action IC0805 "Open European Network for High Performance Computing on Complex Environments", working group WG 4.3: "Numerical Libraries, Solvers & Algorithms" of European Exascale Software Initiative (EESI I) and working group WG 4.1: "Numerical Analysis" of EESI II, managing board of Continuous Optimization Working Group (EUROPT) of the Association of European Operational Research Societies (EURO) and board of Lithuanian Operational Research Society (member society of EURO and IFORS), editorial boards of Central European Journal of Computer Science, Central European Journal of Engineering, Informatica, Journal of Global Optimization, Mathematical Modelling and Analysis, Optimization Letters.