TY - GEN
T1 - An ant-colony-optimization based approach for determination of parameter significance of scientific workflows
AU - Khan, Fakhri Alam
AU - Han, Yuzhang
AU - Pllana, Sabri
AU - Brezany, Peter
PY - 2010
Y1 - 2010
N2 - In the process of a scientific experiment a workflow is executed multiple times using various values of the parameters of activities. For real-world workflows that may contain hundreds of activities, each having several parameters, it is practically not feasible to conduct a parameter sensitivity study by simply following a "brute-force approach" (that is experimental evaluation of all possible cases). We believe that a heuristic-guided approach enables to find a near-optimal solution using a reasonable amount of resources without the need for the evaluation of all possibilities. In this paper we present a novel methodology for determination of parameter significance of scientific workflows that is based on Ant Colony Optimization (ACO). We refer to our methodology, which is a customization of ACO for Parameter Significance determination, as ACO4PS. We use ACO4PS to identify (1) which parameter strongly affects the overall result of the workflow and (2) for which combination of parameter values we obtain the expected result. ACO4PS generates a list of all workflow parameters sorted by significance as well as is capable of generating a subset of significant parameters. We empirically evaluate our methodology using a real-world scientific workflow that deals with the Non-Invasive Glucose Measurement.
AB - In the process of a scientific experiment a workflow is executed multiple times using various values of the parameters of activities. For real-world workflows that may contain hundreds of activities, each having several parameters, it is practically not feasible to conduct a parameter sensitivity study by simply following a "brute-force approach" (that is experimental evaluation of all possible cases). We believe that a heuristic-guided approach enables to find a near-optimal solution using a reasonable amount of resources without the need for the evaluation of all possibilities. In this paper we present a novel methodology for determination of parameter significance of scientific workflows that is based on Ant Colony Optimization (ACO). We refer to our methodology, which is a customization of ACO for Parameter Significance determination, as ACO4PS. We use ACO4PS to identify (1) which parameter strongly affects the overall result of the workflow and (2) for which combination of parameter values we obtain the expected result. ACO4PS generates a list of all workflow parameters sorted by significance as well as is capable of generating a subset of significant parameters. We empirically evaluate our methodology using a real-world scientific workflow that deals with the Non-Invasive Glucose Measurement.
UR - http://www.scopus.com/inward/record.url?scp=77954317928&partnerID=8YFLogxK
U2 - 10.1109/AINA.2010.24
DO - 10.1109/AINA.2010.24
M3 - Conference contribution
AN - SCOPUS:77954317928
SN - 9780769540184
T3 - Proceedings - International Conference on Advanced Information Networking and Applications, AINA
SP - 1241
EP - 1248
BT - 24th IEEE International Conference on Advanced Information Networking and Applications, AINA 2010
ER -