摘要
围绕高维数据降维方法及其应用展开综述和评析,首先对于有监督学习框架下充分降维方法和基于特征选择降维方法的演变、发展进行系统性评述,然后系统介绍在线高维数据降维、无监督学习高维降维和张量降维等最新降维方法的发展脉络,最后列举高维数据降维方法在金融、遥感等领域的应用研究以及亟待解决的问题。
The paper mainly reviewed and analyzed the methods of dimensionality reduction in high-dimensional data,as well as its application.At first,the paper made a systematic review on the evolution and development of the methods of sufficient dimensionality reduction and feature selection for dimensionality reduction under the framework of superised learning.Next,the paper sytematically introduced the rescarch thread of some latest dimensionality reduction methods,such as dimensionality reduction in high-dimensional data,high-dimensional dimensionality reduction in unsupervised learning and tensor dimensionality reduction.At last,the research listed the application and existing problems of the methods of dimensionality reduction in high-dimensional data in the fields of finance and remote sensing.
作者
刘翠霞
史代敏
宋文静
LIU Cui-xia;SHI Dai-min;SONG W en-jing(School of Finance,Chongqing Technology and Business University,Chongqing 400060;School of Statistics,Southwest University of Finance and Economics,Chengdu 611130;School of Finance,University of Intemnational Business and Economics,Beijing 100029,China)
出处
《统计学报》
2023年第3期11-21,共11页
Journal of Statistics
基金
重庆市教委科学技术项目(KJ1600629)
重庆市社会科学规划项目(2015PY25)。
关键词
高维降维
在线降维
张量降维
稀疏降维
无监督学习
high-dimensional reduction
online dimensionality reduction
tensor dimensionality reduction
sparse dimensionality reduction
unsupervsed leaming
作者简介
刘翠霞(1979-),女,宁夏固原人,重庆工商大学金融学院教师,西南财经大学统计学院博士研究生,主要研究方向是金融计量理论及应用、数据资产定价;史代敏(1965-),男,四川泸州人,西南财经大学统计学院教授,经济学博士,博士生导师,主要研究方向是金融投资数量分析;宋文静(2002-),女,河北邯郸人,对外经济贸易大学金融学院硕士研究生,主要研究方向是资本市场投资组合与定价。
