DocumentCode :
2781218
Title :
Measuring algorithm footprints in instance space
Author :
Smith-Miles, Kate ; Tan, Thomas T.
Author_Institution :
Sch. of Math. Sci., Monash Univ., Clayton, VIC, Australia
fYear :
2012
fDate :
10-15 June 2012
Firstpage :
1
Lastpage :
8
Abstract :
This paper proposes a new methodology to determine the relative performance of optimization algorithms across various classes of instances. Rather than reporting performance based on a chosen test set of benchmark instances, we aim to develop metrics for an algorithm´s performance generalized across a diverse set of instances. Instances are summarized by a set of features that correlate with difficulty, and we propose methods for visualizing instances and algorithm performance in this high-dimensional feature space. The footprint of an algorithm is where good performance can be expected, and we propose new metrics to measure the relative size of an algorithm´s footprint in instance space. The methodology is demonstrated using the Traveling Salesman Problem as a case study.
Keywords :
optimisation; travelling salesman problems; algorithm footprint measurement; algorithm performance; high-dimensional feature space; instance space; instance visualization; operations research literature; optimization algorithms; traveling salesman problem; Algorithm design and analysis; Cities and towns; Extraterrestrial measurements; Optimization; Prediction algorithms; Visualization; algorithm footprints; heuristics; instance difficulty; performance metrics; traveling salesman problem;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Evolutionary Computation (CEC), 2012 IEEE Congress on
Conference_Location :
Brisbane, QLD
Print_ISBN :
978-1-4673-1510-4
Electronic_ISBN :
978-1-4673-1508-1
Type :
conf
DOI :
10.1109/CEC.2012.6252992
Filename :
6252992
Link To Document :
بازگشت