位置：成果数据库 > 期刊 > 期刊详情页
基于Beta噪声模型支持向量回归及其应用
  • ISSN号：0469-5097
  • 期刊名称：南京大学学报(自然科学版)
  • 时间：2013.7.30
  • 页码：418-424
  • 分类：O213.2[理学—概率论与数理统计;理学—数学]
  • 作者机构：[1]河北师范大学数学与信息科学学院,石家庄050024, [2]衡水学院数学与计算机学院,衡水053000, [3]天津大学计算机科学与技术学院,天津300072, [4]天津大学软件学院,天津300072
  • 相关基金：国家重点基础研究发展计划（2012CB215201）; 国家自然科学基金项目（61222210,61170107）; 河北省高等学校科学研究计划项目（Z2010188）
  • 相关项目：多源信息系统知识获取的粒计算方法
作者： 张仕光|胡清华|谢宗霞|米据生|
关键词： 支持向量回归, Beta-噪声模型, 损失函数, 不等式约束, 风速预报, support vector regression, beta-noise model, loss function, inequality constraints, wind speed forecasting
中文摘要：

大部分回归技术假设误差服从Gauss分布,并把最小化误差的平方和作为优化目标.然而,风电预报的噪声不服从Gauss分布,而是服从Beta分布.在ν-支持向量回归的基础上,本文提出一种基于不等式约束的Beta噪声支持向量回归（BN-SVR）的新模型,并将其应用于短期风速预报中.实验结果说明,模型BN-SVR是有效可行的,且较经典支持向量回归模型获得更好的性能.

英文摘要：

A lot of regression techniques assume that the errors satisfy the Gaussian distribution and use least-squared errors as the optimization objectives.However,the noise of wind power forecasting does not obey the Gaussian distribution but the Beta distribution.An extension of the support vector regression,called an improved SVR based on Beta-noise model of inequality constraints（BN-SVR）,is proposed in this work.Our experimental results on the real-world data sets of wind speed confirm the validity and feasibility of our BN-SVR,and it obtains better performance than ν-SVR,GN-SVR and HN-SVR.

同期刊论文项目
多源信息系统知识获取的粒计算方法
期刊论文 43 会议论文 1 著作 2
多模态数据知识发现
期刊论文 34 会议论文 1
同项目期刊论文
粗糙ε-支持向量回归模型
Attribute reduction based on maximal rules in decision formal context
The strong direct product of formal contexts
Uncertainty measurement for interval-valued information systems
格观念下的知识约简
粗糙集的两种相似性度量
决策形式背景的两种不确定性度量
Some minimal axiom sets ofrough sets
Computational reasoningbased on complemented distributive lattices
Kernel ridge regression forgeneral noise model with its application
Multi-confidence rule acquisition orientedattribute reduction of covering decision systemsvia combin
Variable precisionmultigranulation decision-theoretic fuzzy rough sets
A Comparative Study of MGRSs and their Uncertainty Measures
信息系统中属性约简的矩阵算法
Attribute Reduction for Heterogeneous Data Based on the Combination of Classical and Fuzzy Rough Set
Evidence-theory-based numerical algorithms of attribute reduction with neighborhood-covering rough s
Structure of feature spaces related to fuzzy similarity relations as kernels
异构形式背景上的不确定推理
基于闭算子的属性约简
A novel variable precision (theta, sigma)-fuzzy rough set model based on fuzzy granules
Belief functions on general intuitionistic fuzzy information systems
决策形式背景下的主观贝叶斯概率推理
基于状态空间模型的中国城镇居民消费结构变迁分析
Exploiting diversity for optimizing margin distribution in ensemble learning
Dynamic classifier ensemble using classification confidence
不协调决策表的不协调度
风险条件下基于收益视角的最优投资决策研究
Generalized Bézier curves and surfaces based on Lupa？ image-analogueof Bernstein operator
一种任意次非均匀 B 样条的细分算法
对偶概念格的属性约简
余等价关系下的粗糙集
Measuring inclusion,similarityand compatibility between intuitionistic fuzzy sets
无限论域中的粗糙近似空间与信任结构
合成集值决策表的属性特征
Rule extraction from support vector machines based on consistent region covering reduction
粗糙ε-支持向量回归模型
Fuzzy information systems and their homomorphisms
Noise model based ν-support vector regression with its application to short-term wind speed forecast
Robust feature selection based on regularized brownboost loss
Robust Object Tracking via Active Feature Selection
Extending a configuration model to find communities in complex networks
A heuristic approach to effective and efficient clustering on uncertain objects
基于半监督距离学习和多模态信息的图像聚类
Feature selection for monotonic classification via maximizing monotonic dependency
An improved attribute reduction scheme with covering based rough sets
Pattern-Based Wind Speed Prediction Based on Generalized Principal Component Analysis
Exploration of classification confidence in ensemble learning
Short-Term Wind speed and Power Forecasting Based on Heteroscedastic Support Vector Regression
Efficient Background Modeling Based on Sparse Representation and Outlier Iterative Removal
Heterogeneous Feature Selection with Multi-Modal Deep Neural Networks and Sparse Group Lasso
Improved support vector machine algorithm for heterogeneous data
Transfer learning for short-term wind speed prediction with deep neural networks
基于邻域粗糙集的多标记分类特征选择算法
Exploring Neighborhood Structures with Neighborhood Rough Sets in Classification Learning
Fuzzy Rough Decision Trees
Comparative analysis on margin based feature selection algorithms
Multi-granularity distance metric learning via neighborhood granule margin maximization
基于多粒度数据压缩的支持向量机
期刊信息
  • 《南京大学学报：自然科学版》
  • 中国科技核心期刊
  • 主管单位:中华人民共和国教育部
  • 主办单位:南京大学
  • 主编：龚昌德
  • 地址：南京汉口路22号南京大学（自然科学版）编辑部
  • 邮编：210093
  • 邮箱：xbnse@netra.nju.edu.cn
  • 电话：025-83592704
  • 国际标准刊号：ISSN：0469-5097
  • 国内统一刊号：ISSN：32-1169/N
  • 邮发代号:28-25
  • 获奖情况:
  • 中国自然科学核心期刊,中国期刊方阵“双效”期刊
  • 国内外数据库收录:
  • 美国化学文摘（网络版）,美国数学评论（网络版）,德国数学文摘,中国中国科技核心期刊,中国北大核心期刊（2004版）,中国北大核心期刊（2008版）,中国北大核心期刊（2011版）,中国北大核心期刊（2014版）,中国北大核心期刊（2000版）
  • 被引量:9316