Yonathan Bard
ABSTRACT
It is desired to find the values of certain system parameters which maximize some performance criterion. Using standard experimental design techniques, one runs an initial set of experiments which explore the system's response surface. Subsequently, the data are smoothed, and a hill climbing technique is used to locate the maximum. This technique was employed successfully to tune the parameters in an experimental version of the VM/370 scheduler.
- 1.M. T. Wasan, "Stochastic Approximation", Cambridge University Press, Cambridge (1969).Google Scholar
- 2.G. E. P. Box and N. R. Draper, "Evolutionary Operation", John Wiley and Sons, New York (1969)Google Scholar
- 3.B. Baumeister and H. Beilner, Stochastic optimization methods and their value for experimenting with computer systems, in "Computer Architectures and Networks", E. Gelenbe and R. Mahl eds., North-Holland, Amsterdam (1974).Google Scholar
- 4.W. G. Cochran and G. M. Cox, "Experimental Designs", John Wiley and Sons, New York (1957).Google Scholar
- 5.Y. Bard, "Nonlinear Parameter Estimation", Academic Press, New York (1974).Google Scholar
- 6.J. Ravin and L. Wheeler, personal communication (1975).Google Scholar
- 7.Y. Bard, Application of the page survival index (PSI) to virtual memory system performance, IBM J. Des. Develop. 19, 212-220 (1975).Google ScholarDigital Library
- 8.F. E. Satterthwaite, Random balance experimentation, Technometrics 1, 111-137 (1959).Google ScholarCross Ref
- 9.Y. Bard, Experimental evaluation of system performance, IBM Systems J. 12, 302-315 (1973).Google ScholarDigital Library
Index Terms
- An experimental approach to system tuning
Recommendations
Fine-Tuning algorithm parameters using the design of experiments approach
LION'05: Proceedings of the 5th international conference on Learning and Intelligent OptimizationOptimizing parameter settings is an important task in algorithm design. Several automated parameter tuning procedures/configurators have been proposed in the literature, most of which work effectively when given a good initial range for the parameter ...
Parameter tuning with Chess Rating System (CRS-Tuning) for meta-heuristic algorithms
Meta-heuristic algorithms should be compared using the best parameter values for all the involved algorithms. However, this is often unrealised despite the existence of several parameter tuning approaches. In order to further popularise tuning, this ...
Experimental Study of Automated Parameter Tuning on the Example of irace and the Traveling Salesman Problem
GECCO '16 Companion: Proceedings of the 2016 on Genetic and Evolutionary Computation Conference CompanionPerformance of optimization algorithms (OA) greatly depends on their parameter values. Automated (offline) parameter tuning methods allow to select good parameter values for an OA. Though such methods are widely used, there is little to none information ...
Comments