What applications can affinity propagation be used for?
It can be applied whenever there is a way to measure or pre-compute a number for each pair of data points, which indicates how similar they are (high values indicate high similarity, while low values indicate low similarity). We have been emailed by people who have used affinity propagation to analyze network traffic, group monkey and human skulls, detect interest points in images, group motion features in videos, identify common trading patters, analyze basketball statistics, identify patterns in audio streams, group together similarly-acting biomolecules and cluster astrophysical objects. Note that similarities need only be provided for data points that can potentially be linked together (see below).
