摘要
根据分块矩阵计算公式和支持向量机核函数矩阵本身特点,在增量式最小二乘支持向量机算法的基础上,通过引入剪枝方法改善最小二乘支持向量机的稀疏性,并将这种方法应用于时间序列预测,试验表明这一方法在预测精度及速度上具有一定的优越性。
By the calculation of block matrix and least squares support vector machine kernel function matrix's property,based on the incremental least squares support vector machine algorithm,the pruning algorithm is introduced to improve the sparseness of the incremental LSSVM in this paper.The algorithm is used to the time series forecasting.Experimental results show that this method is much better than the standard LSSVM algorithm in accuracy and learning speed.
出处
《微型电脑应用》
2009年第6期12-13,6,共3页
Microcomputer Applications
基金
甘肃省自然科学基金(0710RJZA054)
甘肃省有色金属新材料省部共建国家重点实验室开放基金(SKL05010)
关键词
时间序列预测
最小二乘支持向量机
增量式算法
剪枝算法
Time series forecasting
Least squares support vector machine
Incremental algorithm
Pruning algorithm
