Learning for Distributed Systems

Organized by: 

Arnaud Legrand


Panayotis Mertikopoulos

Detailed information: 



How to be soaking wet and still not feel too bad about not taking an umbrella

Starting with the title, this talk will be a hazy excursion in the theory of learning and its applications to large-scale distributed systems. We will start with a brief account of what learning is (hint : more than a "big data" buzzword), what it cannot do (predict the future), and what it can do (make you feel less sorry for consciously leaving your umbrella at home on a rainy day). After presenting some recent advances in the theory (chiefly with regards to learning in games), we will discuss how learning algorithms can be applied to adaptive resource allocation problems in large-scale systems. Our examples will be drawn mostly from the world of communication networks and computing grids, but if there is time, we will also discuss applications to digital signal processing and information theory