Talk:Affinity propagation

From Wikipedia, the free encyclopedia
Jump to: navigation, search
WikiProject Computer science (Rated Start-class, Mid-importance)
WikiProject icon This article is within the scope of WikiProject Computer science, a collaborative effort to improve the coverage of Computer science related articles on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
Start-Class article Start  This article has been rated as Start-Class on the project's quality scale.
 Mid  This article has been rated as Mid-importance on the project's importance scale.

About this similarity function[edit]

I just read in the linked source Article (Brendan J. Frey; Delbert Dueck (2007). "Clustering by passing messages between data points"), that those equations are valid if the similarity function s is the negative square distance of two vectors. I went with the definition in the article and it ended up messing up my script. It is rather important to correct this. I am now only mentioning this for more experienced and confident wikipedions to take a look at.

Similarity example mismatch?[edit]

Where it says, "... that is s(x_i, x_j) > s(x_i, x_k) iff x_j is more similar to x_i than x_k", it seems like the expression is literally saying that x_i is more similar to x_j than x_k. The expression doesn't match the English description of the expression's meaning. Is there a mistake here? Ecashin (talk) 13:16, 25 September 2014 (UTC)

What is meant is: "s(xi, xj) > s(xi, xk) iff the similarity between xi and xj is greater than that between xi than xk", but that seemed rather repetitive. Any suggestions on how to clarify this? QVVERTYVS (hm?) 18:30, 25 September 2014 (UTC)
I just updated the subscripts to match the definition, now that you've confirmed my reading of the text. Thanks! Ecashin (talk) 14:33, 26 September 2014 (UTC)

Quality and importance[edit]

This article is clearly just a stub. Among other details, it needs to explain how one actually determines the clusters bases on a(i,k). Being an algorithm that doesn't require specifying the number of clusters, it's worth having a more thorough article about it. I assigned therefore an importance of Mid level.

There's also no mention of the two important parameters for this algorithm, the preference and the damping factor. The preference parameter directly influences the number of clusters found by the algorithm. However I only have a intuitionally understanding of how the preference works but don't know how exactly it influences the messages or initialisation. — Preceding unsigned comment added by Kugelbrot (talkcontribs) 09:42, 25 August 2016 (UTC)