基于间隔损失和L_1范数调节的特征选择方法研究

Feature Selection Method based on Margin Loss and L_1-norm Regulation

特征评价和选择是机器学习和模式识别的重要步骤。为了获得稀疏特征子集,结合间隔损失评估策略和L1范数调节技术来获得一种有效的特征选择方法(MLFWL-L1),并将其应用到RBFSVM分类器。实验中,在UCI数据集上将提出的算法与Simba和ReliefF对比表明,验证所提出的算法是一种有效的特征选择方法。

Feature selection is an important task in machine learning and pattern recognition.The paper designs an algorithm for sparse feature subset. Then it combines margin loss with Ll-nonn regulation technology to obtain an effective feature selection method （MLFWL-LI） ,and applies it to RBFSVM classifier. Finally, the proposed technique is tested through a series of experiments with UCI data sets. Compared with four methods, the conclusion is that the proposed technique is effective and efficient.