Distributed dimension reduction algorithms for widely dispersed data

It is well known that information retrieval, clustering and visualization can often be improved by reducing the dimensionality of high dimensional data. Classical techniques offer optimality but are much too slow for extremely large databases. The problem becomes harder yet when data are distributed...

وصف كامل

محفوظ في:
التفاصيل البيبلوغرافية
المؤلف الرئيسي: Abu-Khzam, F.N. (author)
مؤلفون آخرون: Samatova, N.F. (author), Ostrouchov, G. (author), Langston, M.A. (author), Al Geist, G. (author)
التنسيق: conferenceObject
منشور في: 2002
الوصول للمادة أونلاين:http://hdl.handle.net/10725/7497
http://libraries.lau.edu.lb/research/laur/terms-of-use/articles.php
http://www.actapress.com/Abstract.aspx?paperId=24561
الوسوم: إضافة وسم
لا توجد وسوم, كن أول من يضع وسما على هذه التسجيلة!
الوصف
الملخص:It is well known that information retrieval, clustering and visualization can often be improved by reducing the dimensionality of high dimensional data. Classical techniques offer optimality but are much too slow for extremely large databases. The problem becomes harder yet when data are distributed across geographically dispersed machines. To address this need, an effective distributed dimension reduction algorithm is developed. Motivated by the success of the serial (non-distributed) FastMap heuristic of Faloutsos and Lin, the distributed method presented here is intended to be fast, accurate and reliable. It runs in linear time and requires very little data transmission. A series of experiments is conducted to gauge how the algorithm’s emphasis on minimal data transmission affects solution quality. Stress function measurements indicate that the distributed algorithm is highly competitive with the original FastMap heuristic.