Title :
A Maximum Entropy Based Scalable Algorithm for Resource Allocation Problems
Author :
Sharma, Puneet ; Salapaka, Srinivasa ; Beck, Carolyn
Author_Institution :
Univ. of Illinois at Urbana Champaign, Urbana
Abstract :
In this paper, we propose a scalable algorithm for solving resource allocation problems on large datasets. This class of problems is posed as a multi-objective optimization problem in a maximum entropy principle framework. This algorithm solves a multi-objective optimization problem that minimizes simultaneously the coverage cost and the computational cost by appropriate recursive prescription of smaller subsets required for a ´divide and conquer´ strategy. It provides characterization of the inherent trade-off between reduction in computation time and the coverage cost. Simulations are presented that show significant improvements in the computational time required for solving the coverage problem while maintaining the coverage costs within pre- specified tolerance limits.
Keywords :
divide and conquer methods; maximum entropy methods; operations research; optimisation; computational cost; coverage cost; divide and conquer strategy; maximum entropy based scalable algorithm; multi-objective optimization; resource allocation; Cities and towns; Computational efficiency; Computational modeling; Cost function; Drugs; Entropy; Libraries; Motion control; Partitioning algorithms; Resource management;
Conference_Titel :
American Control Conference, 2007. ACC '07
Conference_Location :
New York, NY
Print_ISBN :
1-4244-0988-8
Electronic_ISBN :
0743-1619
DOI :
10.1109/ACC.2007.4282846