In this project, the focus of research will be placed on dynamic, high-dimensional and random complex systems and we aim at developing new methodologies for computational inference which are both theoretically sound and practically effective. Hence, we will push the limits of the state of the art in three (certainly non orthogonal) directions:

(i) the theoretical development and analysis of new methodologies for computational inference, specifically devised for the objective class of complex systems;

(ii) the application of the new algorithms and methods in a set of selected applications in the fields of information & communication technology (ICT), bioengineering and environmental sciences; and

(iii) the practical implementation of the most successful methodologies in two real-time testbeds, in order to unequivocally demonstrate the ability to predict and control the behavior of both technological and biological complex systems.

The fundamental theoretical work in (i) is aimed at, first, properly understanding and, second, fully exploiting, the characteristic features of stochastic complex dynamical systems in order to be able to build good mathematical models, validate them against records of real data, _t their parameters accurately or choose among multiple competing models.

Such goals demand significant advancement of the current, state-of-the-art, computational inference methods such as Markov chain Monte Carlo (MCMC), population Monte Carlo (PMC) or particle filters (PFs). There is a plethora of real-world problems involving large-scale systems where adequate computational inference techniques are badly needed. Pursuing (ii), the members of the team will tackle applications in medicine (ECG and EEG analysis), bio-chemistry (stochastic kinetic models), ICT (multi-target tracking, distributed processing in large sensor networks) and environmental sciences (ambient noise modeling and prediction) according to their expertise. Finally, in direction (iii), a major effort will be devoted to the full implementation of two testbeds where the effective application of the proposed methodologies can be demonstrated.