位置：成果数据库 > 期刊 > 期刊详情页
基于聚类技术的集成学习差异性研究
  • ISSN号：2095-2651
  • 期刊名称：《数学研究及应用：英文版》
  • 时间：0
  • 分类：TP181[自动化与计算机技术—控制科学与工程;自动化与计算机技术—控制理论与控制工程]
  • 作者机构：[1]中国人民大学统计学院,北京102218
  • 相关基金：国家自然科学基金重点项目《生物医学中的统计方法研究》（编号：10431010）,教育部重点基地重大项目《空间统计学及其应用研究》（编号：05JJD910001）,中国人民大学应用统计中心资助项目研究成果之一.
作者： 陈凯[1]
关键词： 分类回归树, 自助法, 选择性集成, CART, bootstrap, selective ensemble
中文摘要：

目前，集成学习特别是选择性集成学习研究已经成为统计机器学习研究的一大热点，从众多的个体学习器中选择差异大且效果好的进行集成已被学术界达成共识，但如何度量个体学习器彼此之间的差异性依然是一个难点。本文提出了一种利用变相似度聚类技术来进行选择性集成学习的算法——SE—Bagging Trees算法。模拟数据表明，该算法往往比简单集成学习算法具有更好的学习效果。

英文摘要：

Ensemble learning now becomes much popular in the field of statistical machine learning, and there are many people drawing conclusion that combining those base learning with high diversity and good performance can improve the performance of the total ensemble learning. But how to measure the diversity is still a problem. This paper introduces a new ensemble algorithm, SE-Bagging Trees ensemble algorithm, based on variational similarity cluster technology. Compared with simple learning algorithms, it often produces a better performance.

同期刊论文项目
生物医学中的统计方法研究
期刊论文 100 会议论文 13 获奖 16 著作 5
同项目期刊论文
Estimating cell probabilities under order-restricted odds ratios
Collapsibility of distribution dependence
A project of applied statistical methods in China:review and outlook
统计应用中的陷阱
Bayesian estimation in the multidimensional three-parameter Logistic model
A likelihood-based method for haplotype association studies of case-control data with genotyping unc
Identifiability and estimation of probabilities from multiple databases with incomplete data and sam
Existence, uniqueness and global stability of positive solutions to the food-limited population mode
A novel real time method of signal strength based indoor localization
Comparison between estimates of the hypothetical proportion with and without standardization for a n
基于非对称损失函数的参数设计
Exact Cochran-Armitage trend tests: comparisons under different models
The restricted EM algorithm under linear inequalities in a linear model with missing data
An optimal phase I/II design utilizing severe toxic and moderate toxic response information separate
A new non-randomized model for analysing sensitive questions with binary outcomes
Dynamical convergence tiajectory in networks
Testing for homogeneity of gametic disequilibrium across strata
Maximum likelihood estimates of two-locus recombination fractions under some natural inequality rest
The restricted EM algorithm under inequality restrictions on the parameters
Hidden Markov models with multiple observers. Advanced Intelligent Computing Theories and Applicatio
p300 plays a role in  expression and cell cycle arrest
Structural learning of chain graphs via decomposition
Active learning of causal networks with intervention experiments and optimal designs
A recursive method for structural learning of directed acyclic graphs
Testing?the?equality?of?two?proportions?for combined unilateral and bilateral data
The total chromatic number of Pseudo-Halin graphs with lower degree
Molecular analysis of early rice stamen development using organ-specific gene expression profiling
Stoichiometry and large-scale patterns of leaf carbon and nitrogen in the grassland biomes of China
Structural learning about independence graphs from multiple databases. Advances in Knowledge Discove
纵向研究非随机缺失数据分析
Stepwise method based on confidence bound and information incorporation for identifying the maximum
A generalized confidence interval for the difference of the means of two normal populations
A Note on Nonautonomous Logistic Equation with Random Perturbation
Incorporating genotyping uncertainty in haplotype frequency estimation in pedigree studies
A test of homogeneity of Hardy-Weinberg disequilibrium across strata
Convergence rate of Gibbs sampler and its application
Functional coefficient instrumental variables models
Gibbs样本的收敛速度及其应用
A bottom-up inference of loss rate
Structural learning of graphical models and its applications to traditional Chinese medicine
Identifiability, stratification and minimum variance estimation of causal effects
Decomposition of search for v-structures in DAGs
Assessing phylogenetic relationships of lycium samples using RAPD and entropy theory
Existence, uniqueness, stochastic persistence and global stability of positive solutions of the logi
A Longitudinal study of the effects of family background factors on mathematics achievements using
Acceleration of the EM and ECM algorithms by Aitken \delta^2 method for log-linear models with parti
Learning causal structures based on Markov equivalence class
Minimax confidence bound of the normal mean under an asymmetric loss function
无指导的中医证候诊断数据的隐变量分析
一种改进的K-Prototype聚类算法
流行病学观察性研究中若干不可检验的假定：对照模型和数据缺失机制
Acceleration of the EM algorithm using the vector epsilon algorithm
Testing Linearity for Nonparametric Component of Partially Linear Models
Criteria for surrogate end points
Decomposition of structural learning about directed acyclic graphs
Surrogate Paradox
病例－对照研究中基因型不确定时单倍型关联分析的似然方法
部分线性变系数模型的Profile Lagrange乘子检验
A longitudinal study of the effects of family background factors on mathematics achievements using q
Longitudinal study of the external pressure effects on children's mathematics and science achievemen
Collapsibility for directed acyclic graphs
Some association measurements and their collapsibility
An approach for mining local causal relationships from databases
Optimal algorithms and intuitive explanations for Markowitz's portfolio selection model and Sharpe's
A conjecture on maximum likelihood estimation
The genomic landscapes of histone H3-Lys9 modifications of gene promoter regions and expression prof
具有缺失数据的线性模型的一般线性不等式约束EM算法
Structure exploration and function prediction of SARS coronavirus E protein
Pitman准则下带有半序约束的最大似然估计的优良性
一种利用混合模型研究基因与疾病间关联的方法
基于EM算法约束条件下参数的估计
异方差下利用Bootstrap抽样技巧对基因表达样式进行选择和聚类
基于回归问题的选择性集成算法
两个部分线性模型的比较
基于分类问题的选择性集成学习研究
部分线性变系数模型中误差方差的估计
部分线性模型非参数分量的线性关系检验
基于模糊的投影寻踪算法在中医证候分类中的应用
部分线性模型非参数部分的多项式关系检验
一种改进的集成混合遗传算法
一种基于差异思想的选择性Bagging Trees集成算法研究
一种改进的K-Prototypes聚类算法
中国广电产业发展与地区差异比较——基于Panel data的分析
最值公式在概率统计上的一个应用
正态总体均值与标准差比在简单半序约束下的最大似然估计(英文)
选择性集成在有害藻类治理中的应用
部分线性变系数模型Backfitting估计的渐近性质
常用统计软件查概率统计值的方法
两分类不平衡数据的Boosting算法
部分线性变系数模型中估计的渐近正态性(英文)
期刊信息
  • 《数学研究及应用：英文版》
  • 中国科技核心期刊
  • 主管单位:国家教育部
  • 主办单位:大连理工大学
  • 主编：王仁宏
  • 地址：大连理工大学应用数学系
  • 邮编：116024
  • 邮箱：
  • 电话：0411-84707392
  • 国际标准刊号：ISSN：2095-2651
  • 国内统一刊号：ISSN：21-1579/O1
  • 邮发代号:8-92
  • 获奖情况:
  • 1998年大连市优秀期刊奖,2000年大连市优秀期刊奖
  • 国内外数据库收录:
  • 俄罗斯文摘杂志,美国数学评论（网络版）,德国数学文摘,日本日本科学技术振兴机构数据库,中国中国科技核心期刊
  • 被引量:36