| 一种基于支持向量回归方法在RoboCup中的应用 |
| |
| 引用本文： | 刘扬,王浩,方宝富,姚宏亮.一种基于支持向量回归方法在RoboCup中的应用[J].合肥工业大学学报(自然科学版),2007,30(10). |
| |
| 作者姓名： | 刘扬 王浩 方宝富 姚宏亮 |
| |
| 作者单位： | 合肥工业大学,计算机与信息学院,安徽,合肥,230009 |
| |
| 基金项目： | 合肥工业大学科学研究发展基金资助项目(050506fF)，安徽省自然科学基金资助项目(070412064) |
| |
| 摘 要： | 对RoboCup中截球问题引入了支持向量回归方法,通过采集样本训练预测模型来预测Agent成功截到球时球运动过的距离。为了达到更好的预测效果,对此模型的参数选择问题进行了相应的研究,最后将此预测模型与广义回归神经网络等传统方法相比较,结果表明,在截球距离的预测精度方面要优于传统的广义回归神经网络。
|
| 关 键 词： | 支持向量回归 径向基网络 预测 |
Application of the method of support vector regression in RoboCup |
| |
| LIU Yang,WANG Hao,FANG Bao-fu,YAO Hong-liang.Application of the method of support vector regression in RoboCup[J].Journal of Hefei University of Technology(Natural Science),2007,30(10). |
| |
| Authors: | LIU Yang WANG Hao FANG Bao-fu YAO Hong-liang |
| |
| Abstract: | The method of support vector regression(SVR) is applied to solving the problem of ball-interception in RoboCup.The SVR algorithm is described,and simulation experiment is carried out.The distance that the ball covers when the agent successfully intercepts the ball is predicted through the prediction model which is based on the collected examples.The optimal learning parameters are determined through cross validation.Simulation results show that the presented prediction model is superior to the artificial neural networks in the precision of prediction. |
| |
| Keywords: | support vector regression radial basis network prediction |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
