Joseph Sifakis - Une vision pour l’Informatique - La perspective système

14:00
Jeudi
3
Nov
2011
Organisé par : 

Laurent Besacier

Intervenant : 

Joseph Sifakis


Information détaillée : 

Joseph Sifakis est ingénieur électricien de l’École Polytechnique d’Athènes, docteur-ingénieur de l’Université scientifique et médicale de Grenoble (USMG)[2] et docteur d’État en informatique de l’USMG et de l’Institut national polytechnique de Grenoble.

Il a fondé le laboratoire Verimag de Grenoble où il travaille encore à présent. Il est le lauréat, avec Edmund Clarke (Carnegie Mellon University) et Allen Emerson (Université du Texas à Austin) du Prix Turing 2007, et le premier Français à recevoir cette distinction. Il a également reçu en 2001 la médaille d’argent du CNRS.

Il est l’un des créateurs de la méthode d’énumération et de vérification de modèles (model checking), pour laquelle il a reçu ce prix. Il s’est également illustré dans l’étude des systèmes hybrides.

Résumé : 

In this talk, I will discuss the evolution of Computer Science and in particular its shift of focus from algorithms and programs to systems. I will advocate for a coherent scientific foundation of system design and present a vision for its development in three work directions :

  • Marrying Physicality and Computation : Computation models ignore physical time and resources and are by their nature very different from analytic models used in physical systems engineering. In order to take into account interaction of computing systems with physical environments they must be enriched and extended with paradigms and methods from Electrical Engineering and Control Theory.
  • Component-based Construction : Complex systems are designed by assembling heterogeneous components. Heterogeneity has different sources including a large variety of interaction mechanisms, synchronous or asynchronous execution and different levels of abstraction. We need theoretical frameworks supporting meaningful and natural composition of heterogeneous components which is essential for tractable and productive system design.
  • Adaptivity : Complex systems must provide a service meeting given requirements in interaction with uncertain environments. It is impossible to predict at design time by case analysis all the potentially critical situations. Adaptivity is a means to enforce correctness in the presence of uncertainty by using control-based techniques. It encompasses a new and realistic vision for “intelligent systems” quite different from the “strong” vision of Artificial Intelligence.

I will conclude with general remarks about the nature of Computer Science as a scientific discipline on its own right and advocate for a deeper interaction and cross-fertilization with other more mature disciplines.