Design and Application of Support Vector Regression Algorithm Based on Ant Colony Optimization

• Published in: 2009 International Conference on Computational Intelligence and Natural Computing Vol. 2; pp. 182 - 185
• Main Authors: Han Zhen-yu, Lian Ming, Fu Hong-ya
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.06.2009
• Subjects: Algorithm design and analysis; Ant colony optimization; Error compensation; Kernel; Linearity; Mean square error methods; non-linear fitting; Pattern recognition; Polynomials; Space technology; Support vector machines; support vector regression
• ISBN: 9780769536453; 076953645X
• DOI: 10.1109/CINC.2009.69

A new data fitting algorithm based on ant colony optimization (ACO) and support vector regression (SVR) is proposed. Ant colony algorithm optimizes three parameters of SVR, including penalty parameter C, insensitive loss function epsiv and kernel function sigma. SVR constructs hyperplane in high dimension space and fits the data in non-linear form. Mean square error of fitting result is used as target of ant colony optimization. ACO finds the best parameters which correspond to the least mean square error. Then, build error compensation scale according to the prediction result, store the scale in motion control card and compensate error in real time. At last, compared ACO-SVR fitting algorithm with polynomial fitting and cubic spline fitting, the results showed that peak-peak value after ACO-SVR compensation was improved from 12.1" to 2.3", which was superior to polynomial fitting (4.95") and cubic spline fitting (2.85"). The ACO-SVR compensation algorithm was proved availability and it was used to compensate shafting system of simulator table successfully.