摘要
针对传统分类算法在葡萄酒品质分类中,对少样本类识别率低的问题,提出一种基于支持向量机的集成学习(AdaBoost-SVM)分类方法。利用网格搜索法对支持向量机基分类器进行建模参数优化;通过构建AdaBoost-SVM集成学习方法将多个基分类器集成,建立以多分类器优化集成为核心的品质分类模型。以UCI数据库中的Wine Quality数据集为研究对象,进行葡萄酒品质分类建模,仿真结果表明,与标准的SVM算法相比,AdaBoost-SVM方法有效提高了少样本类分类正确率、整体样本的分类精度以及泛化率。
Aiming at the problems of the traditional classification algorithms for wine quality classification like low recognition rate for minority-classs amples,an ensemble learning classification algorithm based on support vector machine (SVM) was pro-posed The parameters of SVM base classifier were optimized using the grid search method. The multi base classifiers were in-tegrated by constructing AdaBoost-SVM ensemble learning methods? and the classification model of wines quality was estab-lished using multi-classifiers optimal integration as the core. The classification model was established based on the wine quality datasets of UCI database. The simulation results show that compared with the standard SVM algorithm, the AdaBoost-SVM method not only improves the classification accuracy of minority-class examples, but also improves the classification accuracy and generalization rate of the sample.
出处
《计算机工程与设计》
北大核心
2017年第9期2541-2545,共5页
Computer Engineering and Design
基金
陕西省科技厅科学技术研究发展计划基金项目(2014K15-03-06)
西安市科技计划基金项目(NC1403(2)
NC1319(1))
陕西省社会发展科技攻关基金项目(2015SF277)
关键词
分类
支持向量机
集成学习
葡萄酒品质
不平衡数据
classification
support vector machine
ensemble learning
wine quality
imbalanced data