论文部分内容阅读
在传统SVM的分类求解算法中,由于严格凸的无约束最优化问题中单变量函数x+是不可微的,不能使用通常的最优化的算法进行求解。三次Hermite插值多项式光滑的支持向量机模型采用的是一种多项式光滑技术,用三次Hermite插值多项式代替单变量函数x+,将原来不可微的模型变为可微的模型,并且给出了三次Hermite插值多项式光滑化单变量函数x+的推导过程。使用UCI机器学习数据集中的数据,通过实验验证了该模型的有效性。
In the traditional SVM classification algorithm, the univariate function x + in a strictly convex unconstrained optimization problem is not differentiable and can not be solved by the usual optimization algorithm. Third Hermite Interpolation Polynomial Smoothing Support Vector Machine Model uses a polynomial smoothing technique that uses cubic Hermite interpolation polynomials instead of the univariate function x + to make the original non-differentiable model into a differentiable model and gives three Hermite Interpolations The Process of Deriving Polynomial Smoothing Univariate Function x +. UCI machine learning data using the data set, the effectiveness of the model verified by experiments.