机器学习中的最优化模型与对偶  

Optimization Models and Dualities Study in Machine Learning

作　　者：刘红英[1] 杨茜[1] 

机构地区：[1]北京航空航天大学数学与系统科学学院,北京100083

出　　处：《数学的实践与认识》2017年第4期225-233,共9页Mathematics in Practice and Theory

基　　金：国家自然科学基金(61172060;61403011)

摘　　要：数学最优化是以数学的方式来刻画和找出问题最优解的一门学科.机器学习利用数据构造预测方法,并对这些方法进行研究.介绍了机器学习中与支持向量机和稀疏重构相关的最优化模型.在此基础上,给出了三个典型最优化模型的对偶问题,并详细地讨论了对偶在求解这些问题中的应用.The mathematical optimization is a subject which focuses on characterizing and finding the optimal solution of the practical problem. Machine learning explores the study and construction of algorithms that can learn from and make predictions on data. In this paper, the support vector machine and sparse reconstruction in machine learning field are introduced which are typical optimization models. The duality with application to these optimization problems is discussed.

关 键 词：对偶 机器学习 支持向量机 稀疏重构 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]