论文部分内容阅读
针对v-支持向量机(v-SVM)用于大规模、多峰样本建模时易出现训练速度慢和回归精度低的问题,提出基于边界向量提取的多尺度v-SVM建模方法.该方法采用一种自适应边界向量提取算法,从训练样本中预提取出包含全部支持向量的边界向量集,以缩减训练样本规模,并通过求解多尺度v-SVM二次规划问题获取全局最优回归模型,从多个尺度上对复杂分布样本进行逼近.仿真结果表明,基于边界向量提取的多尺度v-SVM比v-SVM具有更好的回归结果.
Aiming at the problem of v-SVM being prone to slow training and low regression accuracy when modeling large-scale and multi-peak samples, a multi-scale v-SVM modeling method based on boundary vector extraction is proposed. Methods An adaptive boundary vector extraction algorithm was used to pre-extract the boundary vector set containing all the support vectors from the training samples to reduce the training sample size and obtain the global optimal regression by solving the multiscale v-SVM quadratic programming problem Model is proposed to approximate complex distribution samples from multiple scales.The simulation results show that the multi-scale v-SVM based on boundary vector extraction has better regression results than v-SVM.