Publication View

Screening the Parameters Affecting Heuristic Performance (2007)

Abstract
This research screens the tuning parameters of a combinatorial optimization heuristic. Specifically, it presents a Design of Experiments (DOE) approach that uses a Fractional Factorial Design to screen the tuning parameters of Ant Colony System (ACS) for the Travelling Salesperson problem. Screening is a preliminary step to building a Response Surface Model (RSM) [20, 18]. It identifies those parameters that need not be included in a Response Surface Model, thus reducing the complexity and expense of the RSM design. 10 algorithm parameters and 2 problem characteristics are considered. Open questions on the effect of 3 parameters on performance are answered. Ant placement and choice of ant for pheromone update have no effect. However, the choice of parallel or sequential solution construction does indeed influence performance. A further parameter, sometimes assumed important, was shown to have no effect on performance. A new problem characteristic that effects performance was identified. The importance of measuring solution time was highlighted by helping identify the prohibitive cost of non-integer parameters where those parameters are exponents in the ACS algorithm’s computations. All results are obtained with a publicly available algorithm and problem generator.

Publication details
Download http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.65.1349
Source http://www-users.cs.york.ac.uk/~eridge/publications/conferences/GECCO/2007_GECCO_2a.pdf
Publisher ACM
Contributors CiteSeerX
Repository CiteSeerX - Scientific Literature Digital Library and Search Engine (United States)
Keywords Design of Experiments, fractional factorial design, parameter screening, Ant Colony Optimization, Travelling Salesperson Problem
Type text
Language English
Relation 10.1.1.97.3555, 10.1.1.21.6368, 10.1.1.25.1550, 10.1.1.14.9584, 10.1.1.18.4764, 10.1.1.26.2382, 10.1.1.96.6751, 10.1.1.64.3468, 10.1.1.64.7730, 10.1.1.103.4634, 10.1.1.63.3692, 10.1.1.111.4332