Approximation Algorithms for Clustering Points in the Distributed Model

Please enable javascript in your browser.

Aghajani, Emad | 2016

814 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 48596 (19)
University: Sharif University of Technology
Department: Computer Engineering
Advisor(s): Zarrabi Zadeh, Hamid
Abstract:
Clustering is one of the most well-known fundamental problems in computer science. In this thesis we have focused on a particular version of such problem, called capacititated k-center, and we analyze and survey them in the distributed model, when massive data is given. Our contribution in this research includes a new approximation algorithms with constant approximate factors for such problems in the distributed model, as well as improving the best available algorithm for capacitated k-center problem. Composable coreset as one of the most important techniques in distributed and streaming model is our primary tools in designing these algorithms. This technique gives the opportunity of solving the problem in smaller chunks of data, and giving the result by combining them
Keywords:
Distributed Electric Model ; Clustering ; Approximate Algorithm ; Composable Core Sets ; Capacitated K-Center