Practical applications of affinity propagation I am learning about machine learning here.
They took a set of prices for specific companies on the stock market and graphed them:

I would like to know what are some practical applications of affinity propagation?
Besides the computer science space, would there be any other real-world examples?  
 A: The University of Toronto has a page on affinity propogation which lists a bunch of useful examples that are relevant outside the realms of computer science, including:

  
*
  
*How to identify a small number of face images that together accurately represent a data set of face images
  
*How to identify a small number of sentences that accurately reflect the content of a document
  
*How to identify a small number of cities that are most easily accessible from all other cities by commercial airline
  
*How to identify segments of DNA that reflect the expression properties of genes
  

I'm not familiar with the area myself, but I can really see something like the movie shown on that page being used to good effect in a sci-fi film: plot all your enemies, then use affinity propagation to pin-point which ones to take out first. Could make that look really flashy :)
In general, I guess it would be useful for identifying the Lynchpin in any complex structure.
A: One practical thing that one would note while performing data clustering is that, in many occasions, the distance (or similarity) between the data instances will not be symmetric. That is,

Distance (i, j) != Distance (j,i), where i,j belong to {Data Instances} and i != j

Or even sometimes,

Distance (i,i) != 0

Also in some occassions, the distance measures do not satisfy triangle inequality.
These type of measures are called non metric distance measures (as they do not satisfy metric axioms).
More information on such non metric could be found here.
Many well-known clustering algorithms like K-means, Hierarchical Agglomerative clustering, EM etc. were originally designed to operate on metric distances (some variations of such algorithms work on non metric distances as well). One area where Affinity Propagation (AP) truly stands out is that, AP by design can handle non metric measures! 
Excerpts from AP paper in science 2007:  

Unlike metric-space clustering techniques such as k-means clustering,
  affinity propagation can be applied to problems where the data do not
  lie in a continuous space. Indeed, it can be applied to problems where
  the similarities are not symmetric  and to problems where the
  similarities do not satisfy the triangle inequality.

