OPTIMISATION ET APPRENTISSAGE
Objectives
The objective of this course is to give an overview of the methods and formal tools related to the design of adaptive and/or distributed algorithms, as well as to the optimization of the parameters of a telecommunication system or in networks, generally using a learning function of the system state.
In this context, several points will be addressed:
1/Adaptive and distributed algorithms: LMS/RLS algorithms, stochastic gradient based methods. Application to adaptive signal processing and sensor networks.
2/Optimization for telecommunications: nonlinear programming with constraints, convex programming, dynamic programming, heuristic methods: application to resource allocation and scheduling;
3/Detection, Classification and Learning: blind detection, classification principles in telecommunications, neural learning methods, reinforcement learning and decision markov processes.
4/Modeling
Bibliography
VHDL - langage, modélisation, synthèse (R. AIRIAU et al. - Presses Polytechniques et Universitaires Romandes)