Talk:Stress majorization

Speedy deletion
Stress majorization is a very old (1977) and widely used technique for metric multidimensional scaling with a number of other interesting applications. It might take me a few days to put it all together though, please don't delete it! It may be a little stub-like at the moment but I promise to keep expanding it as time permits. Thanks Tgdwyer 07:37, 26 March 2007 (UTC)


 * Sorry I was a bit hasty on the speedy delete tag. Still, you need to make this a bit less technical so average readers (or below-average readers such as myself) might have a faint clue about this topic. A real-world example might be in order. I realize you'll never dumb it down to an eighth-grade level, and you wouldn't really want to. But still, just a little context for the rest of us would be a good idea. Realkyhick 05:24, 27 March 2007 (UTC)

X'
I can't make out what $$X'$$ is referring to here. $$X$$ is a configuration matrix. Is $$X'$$ just the transpose of $$X$$? Cai (talk) 16:30, 24 September 2009 (UTC)

Majorization or minimisation?
I expect the description is technically correct but it does read in a bit of an odd way:

"Stress majorization is an optimization strategy used in multidimensional scaling (MDS) where, for a set of n, m-dimensional data items, a configuration X of n points in r(<<m)-dimensional space is sought that minimises the so called stress function σ(X). "

This seems to be saying that "majorization" (which I suspect is not a real word) consists of "minimisation" which sounds very odd to me when I read it. Is there a better way of wording this?90.205.123.105 (talk) 18:36, 9 April 2011 (UTC)

The SMACOF algorithm isn't clear
It states "i.e. for the Hessian matrix V the second term is equivalent to tr X'VX "

What Hessian matrix V? Hessian of which function? — Preceding unsigned comment added by Uriamor (talk • contribs) 19:39, 31 December 2017 (UTC)