CLASS-L Archives

June 2001


Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Mel Janowitz <[log in to unmask]>
Reply To:
Classification, clustering, and phylogeny estimation
Thu, 21 Jun 2001 11:49:45 -0400
text/plain (78 lines)
 Subject: DIMACS Workshop on Algorithms for Multidimensional Scaling

 DIMACS Workshop on Algorithms for Multidimensional Scaling

 Date:  August 10, 2001
 Location:  DIMACS Center, CoRE Building, Rutgers University


 Douglas Carroll, Rutgers University, [log in to unmask]
 Phipps Arabie, Rutgers University, [log in to unmask]

 Presented under the auspices of the Special Focus on
 Data Analysis and Mining.

 Talks will primarily focus on algorithmic aspects of MDS.
 Primary emphasis will be given to use of algorithms not
 typically used for fitting MDS models, such as linear and
 mixed integer programming, nonlinear or dynamic programming,
 and other optimization methods that have not been traditionally
 applied to fitting MDS and related models-- especially when
 these can be used to fit models not easily amenable to more
 traditional optimization techniques, such as various gradient
 based procedures. Another class of algorithmic issues to be
 considered has to do with approaches for increasing the size
 of data sets MDS and related methods can deal with, either via
 improvements in existing algorithms aimed at speeding them
 up considerably, as well as enabling them to deal with larger
 data sets, or by use of heuristic methods that may not precisely
 optimize a well defined criterion of fit, but may allow dealing
 with much larger data sets efficiently. Papers of this type can
 generically be classified as papers on MDS and related techniques
 for Massive Data Sets (MDS for MDS), or "Data Mining" in the
 context of MDS and related methodology.

 The class of MDS and related models that will be dealt with include,
 in addition to spatial models for proximity data, multidimensional
 or multiattribute models for preferential choice or other multivariate
 data, non-spatial or discrete models, such as tree structure,
 (overlapping or non-overlapping) clustering models, or "hybrid"
 models combining aspects of continuous spatial and discrete
 non-spatial models (e.g., a model for proximity data in which
 proximities are related to a sum of distances from an MDS-like
 spatial model and an ultrametric or path length metric defined on
 one or more tree structures; alternatively, the discrete component
 could consist of distances or distance-like measures defined on
 pairs of objects based on, say, an overlapping clustering structure).

 Our funding agencies require that we charge a registration fee during
 the course of the workshop.  Registration fees include participation
 in the workshop, all workshop materials, breakfast, lunch, breaks,
 and any scheduled social event (if applicable).  Fees are $40 per
 person for faculty, researchers and "other", and $5 per person for
 postdocs. However, the registration fee is waived for undergraduate
 students, graduate students, DIMACS postdocs and DIMACS long-term
 visitors who are in residence at DIMACS.  Registration fee to be
 collected on site, cash, check, VISA/Mastercard only.  Fees for
 all DIMACS Members are covered through their institution's membership
 in DIMACS and therefore no DIMACS Member needs to pay the registration fee.

 Information on participation, registration, accommodations, and travel
 can be found at:

* Melvin F Janowitz                                     *
* 404 CoRE Building     Phone: 732-445-5931             *
* DIMACS                email: [log in to unmask]  *
* Rutgers, The State University                         *
* 96 Frelinghuysen Road                                 *
* Piscataway, NJ 08854                                  *
*                                                       *