Abstract
For metric multidimensional scaling much attention is given to algorithms for computing the configuration for fixed dissimilarities. Here we study the inverse problem: what is the set of dissimilarity matrices that yield a given configuration as a stationary point? Characterizations of this set are given for stationary points, local minima, and for full-dimensional scaling. A method foi computing the inverse map for stationary points is presented along with several examples.
| Original language | English |
|---|---|
| Pages (from-to) | 3-21 |
| Number of pages | 19 |
| Journal | Journal of Classification |
| Volume | 14 |
| Issue number | 1 |
| DOIs | |
| Publication status | Published - 1997 |