Document Type
Conference Proceeding
Conference
ANNIE 2009, ANNs in Engineering
Publication Date
2009
City
St. Louis, Missouri
Abstract
This paper explores the feasibility of applying support vector regression (SVR) kernel-based supervised learning method to develop hot mix asphalt (HMA) dynamic modulus (|E*|) predictive models. SVR-based prediction models were developed using the latest comprehensive |E*| database that is available to the researchers. The SVR model predictions were compared with the existing regression-based prediction model which is employed in the Mechanistic-Empirical Pavement Design Guide (MEPDG). The SVR based |E*| models show better prediction accuracy compared to the existing regression models. The determination of optimal function and parameters for SVR algorithm is recommended to improve the prediction performance of SVR based |E*| models.
Copyright Owner
Gopalakrishnan et al
Copyright Date
2009
Language
en
Recommended Citation
Gopalakrishnan, Kasthurirangan; Kim, Sunghwan; and Ceylan, Halil, "Hot Mix Asphalt Dynamic Modulus Prediction Using Kernel Machines" (2009). Civil, Construction and Environmental Engineering Conference Presentations and Proceedings. 13.
https://lib.dr.iastate.edu/ccee_conf/13
Comments
This is a manuscript of an article from ANNIE 2009, ANNs in Engineering, St. Louis, Missouri, November 2-4, pp. 131-138. Posted with permission.