|Title:||On convex formulation of 1D scaling problem
|Abstract:||The 1D-scaling problem is to locate a set of points on an axis using [noisy] matrix of pair-wise distances between them. Straight-forward least-squares formulation of this problem is non-convex, and there is a danger of being trapped in a local minimum. We propose a convex variational formulation, which provides reliable solution to the prob- lem. Noise statistics can be incorporated into the objective in quasi-optimal way. We also introduce a parametric family of objective functions, which gradually transforms from convex to "statistically optimal" non-convex. Using sequential optimizations with gradually updated "non-convexity" parameter, further improvement of solution can be achieved. A priori knowledge about the order of point locations can be incorporated as a set of linear constraints.|
|Copyright||The above paper is copyright by the Technion, Author(s), or others. Please contact the author(s) for more information|
Remark: Any link to this technical report should be to this page (http://www.cs.technion.ac.il/users/wwwb/cgi-bin/tr-info.cgi/2008/CS/CS-2008-12), rather than to the URL of the PDF files directly. The latter URLs may change without notice.
To the list of the CS technical reports of 2008
To the main CS technical reports page
Computer science department, Technion