Combinations of adaptive filters: low-cost solutions and adaptive networks

Abstract

This visit aims to continue the ongoing collaboration between our groups at Universidad Carlos III and USP, in the general subject of cooperative estimation, particularly combinations of adaptive filters and adaptive networks. Regarding combinations of adaptive filters, our groups have published several conference papers. We intend to complete work on a theoretical model for convex combinations of adaptive filters, and on methods to reduce the complexity of the combination, with the goal of preparing journal papers. With respect to adaptive networks, we have been working on methods to adaptively choose the weights used in the exchange of information between the nodes, and on strategies for network estimation when the parameter vectors that each node is trying to estimate are not all equal, and intend also to complete new papers on the subject during the visit. Finally, one of Prof. Arenas' areas of expertise is machine learning, a subject in which adaptive filtering algorithms have started to be applied recently. A last goal of the visit is for us to study the possibility of future collaboration in this area. (AU)