基于线性孪生支持向量机的特征选择方法

Feature selection method based on linear Twin Support Vector Machine

提出了一种基于线性孪生支持向量机(TWSVM)的嵌入式特征选择方法。该方法在构造分类器的过程中,通过在TWSVM原有优化模型中引入一个惩罚项,来实现特征选择。在求解过程中,采用交替迭代优化方法将该模型求解问题分解成两个子问题来处理,即标准TWSVM优化问题和关于特征权重的非线性约束优化问题,并分别对子问题进行有效求解。在UCI数据集上对算法进行了仿真分析和比较,仿真结果验证了算法的有效性。

A new embedded feature selection method based on linear Twin Support Vector Machine(TWSVM)is proposed.It selects features during classifier construction by introducing a penalty term in the primal formulation of TwinSupport Vector Machine. In the solving process, it utilizes alternating iterative optimization method to decompose theproblem of solving the model into two sub-problems, namely the standard TWSVM optimization problem and the nonlinearconstrained optimization problem about feature weight, and effectively solves the sub-problems respectively. The featureselection method is analyzed and compared on UCI datasets. Simulation results verify the proposed method is effective.