摘要
现有的序数回归方法能够利用先验的有序信息获得比一般多分类方法更加优越的性能,但忽略了仪器不精密和环境扰动等因素产生的复杂不确定性数据。针对序数回归中数据存在的不确定信息问题,基于最大间隔原理,构建面向不确定数据的支持向量序数回归模型。实验结果表明,模型可以减少不确定数据对决策边界的影响,增强学习模型的鲁棒性。
Ordinal regression can predict ordered multi-class results for discrete and ordered data.Although the existing ordinal regression methods can use the prior ordered information to obtain better performance than the general multi-class methods,they ignore the complex uncertain data generated by instrument imprecision and environmental disturbance.To solve the problem of uncertain data in ordinal regression,a support vector ordinal regression model for uncertain data was constructed based on the principle of maximum interval.Experimental results show that the model can reduce the influence of uncertain data on decision boundary and enhance the robustness of the learning model.
作者
李晰
肖燕珊
刘波
LI Xi;XIAO Yan-shan;LIU Bo(School of Computer,Guangdong University of Technology,Guangzhou 510006,China;School of Automation,Guangdong University of Technology,Guangzhou 510006,China)
出处
《计算机工程与设计》
北大核心
2023年第1期174-181,共8页
Computer Engineering and Design
基金
国家自然科学基金项目(61672169)。
关键词
机器学习
多分类学习
不确定数据
序数回归
有序信息
最大间隔原理
支持向量机
machine learning
multi-class learning
uncertain data
ordinal regression
ordered data
maximum interval
support vector machine
