Zhao, XiaochuanSayed, Ali H.2017-12-192017-12-192017-12-19201310.1109/ICASSP.2013.6638657https://infoscience.epfl.ch/handle/20.500.14299/143348This work shows how the combination weights of diffusion strategies for adaptation and learning over networks can be chosen in order for the network mean-square-error performance to match that of an optimized centralized (or batch) solution. The results show that this is possible regardless of the network topology, however sparse it is, as long as the network is connected without disjoint sub-graphs.text::conference output::conference proceedings::conference paper