支持向量机在物理实验中的应用

Applications of Support Vector Machine in Physics Experiment

回归型支持向量机方法SVR具有很好的学习性能。本文结合两个物理实验提出了利用SVR方法对实验数据进行曲线拟合,并与最小二乘法的方法进行了比较。实验表明其在精度上优于最小二乘法的方法,在对复杂曲线拟合时效果尤为明显。

Support Vector Machine for regression (SVR) has shown very good learning performance. By using the SVR, this article presents a method to curvefit the experiment data in two physics experiments. The experiments show that the proposed SVR-based method is better than the least square algorithms based method in accuracy, especially for complex curvefitting.