Portfolio多目标候选决策方案的M-PBIL方法
详细信息 查看官网全文
- 英文论文题名：M-PBIL Method for the Multiobjective Candidate Solution of Portfolio
- 论文作者：胡仕成 ; 刘腾娇 ; 何青松 ; 考永贵 ; 王嵩
- 英文论文作者：HU Shi-cheng ; LIU Teng-jiao ; HE Qinrsong ; KAO Yong-gui ; WANG Song ; School of Economics & Management ; Harbin Institute of Technology ; School of Science ; Harbin Institute of Technology
- 年：2016
- 作者机构：哈尔滨工业大学经管学院;哈尔滨工业大学理学院;
- 论文关键词：Portfolio ; 多目标优化算法 ; 进化算法 ; 交互式决策
- 英文论文关键词：portfolio ; multiobjective optimization algorithm ; evolutionary algorithm ; interactive decision
- 会议召开时间：2016-11-12
- 会议录名称：第十八届中国管理科学学术年会论文集
- 英文会议录名称：Proceedings of the 18th Chinese Annual Conference on Management Science
- 语种：中文
- 分类号：C934
- 学会代码：ZHYJ
- 会议名称：第十八届中国管理科学学术年会
- 会议地点：中国陕西西安
- 主办单位：中国优选法统筹法与经济数学研究会、西安交通大学、中国科学院科技战略咨询研究院、《中国管理科学》编辑部
- 学会名称：中国优选法统筹法与经济数学研究会
- 页数：6
- 文件大小：444k
- 原文格式：O
- 会议级别：全国
摘要
多目标优化问题Portfolio是投资管理中的决策难题,交互式决策是帮助决策者获得满意投资方案的一种重要途径。针对在交互式决策过程中如何获取Portfolio的多目标候选决策方案,提出了一种求解Portfolio的多目标优化方法M-PBIL。和传统的基于现有解重组生成个体的进化算法不同,M-PBIL遵循了PBIL基于慨率模型生成个体的策略,探索了求解多目标优化问题的三个关键技术:首先,对于连续空间的多目标优化问题Portfolio,提出了实数值的编码方案,克服了传统PBIL二进制编码的编码冗余和概率冲突问题;其次,设计了基因位的变长概率模型,实现了决策变量取值区间的动态划分;再者,对进化过程中的个体采用了基于支配性和代表性的评价机制,解决了多目标优化问题非劣解的选择问题。对M-PBIL的求解性能从收敛性和分布性两个方面采用了定性和定量的评价,在标准数据集上和有代表性的NSGAII进行了比较。实验结果充分表明,M-PBIL具有较好的收敛性和分布性。
Multiobjective optimization problem Portfolio is a hard decision making problem in investment management.Interactive decision making is an important approach to help the decision maker get his satisfied investment solution for Portfolio.With respect to how to obtain the multiobjective candidate decisive solutions for Portfolio in the interactive decision making,a multiobjective optimization method M-PBIL to Portfolio is proposed.Different from the traditional evolutionary algorithms which generate individuals based on the recombination of the current ones,M-PBIL follows the strategy of PBIL to generate individuals based on probability model and investigates three key technologies in solving multiobjective optimization problems.First,for the multiobjective optimization problem in continuous space,a real umber based coding scheme is proposed,which can overcome the defects of binary coding such as code redundancy and probability conflict.In the second place,a variable probability model for the gene bit is designed so as to realize the dynamic partition for the intervals of decision variables.Next,a dominance and representativeness based assessment mechanic is employed for the selection of non-dominated solutions of multiobjective optimization problem.The performances of the M-PBIL are evaluated by convergence and distribution and compared with the representative NSGAII on benchmark data.The experimental results show that M-PBIL outperforms NSGAII in convergence and distribution.
Multiobjective optimization problem Portfolio is a hard decision making problem in investment management.Interactive decision making is an important approach to help the decision maker get his satisfied investment solution for Portfolio.With respect to how to obtain the multiobjective candidate decisive solutions for Portfolio in the interactive decision making,a multiobjective optimization method M-PBIL to Portfolio is proposed.Different from the traditional evolutionary algorithms which generate individuals based on the recombination of the current ones,M-PBIL follows the strategy of PBIL to generate individuals based on probability model and investigates three key technologies in solving multiobjective optimization problems.First,for the multiobjective optimization problem in continuous space,a real umber based coding scheme is proposed,which can overcome the defects of binary coding such as code redundancy and probability conflict.In the second place,a variable probability model for the gene bit is designed so as to realize the dynamic partition for the intervals of decision variables.Next,a dominance and representativeness based assessment mechanic is employed for the selection of non-dominated solutions of multiobjective optimization problem.The performances of the M-PBIL are evaluated by convergence and distribution and compared with the representative NSGAII on benchmark data.The experimental results show that M-PBIL outperforms NSGAII in convergence and distribution.
引文
[1]Khin T.L.Evolutionary approaches for portfolio optimization[D].Nottinham:University of Nottingham,2015.
[2]何朝林.均值—方差模型具有一般不确定性下的最优资产组合选择[J].中国管理科学,2015,23(12):63-70.
[3]苏治,秦磊,方彤.含有图结构约束的稀疏最小方差资产组合模型[J].中国管理科学,2015,23(9):65-70.
[4]Meignan D,Knust S,Frayret J M,et al.A review and taxonomy of interactive optimization methods in operations research[J].ACM Transactions on Interactive Intelligent Systems,2015,5(3):1-43.
[5]Deb K.Pratap A,Agarwal S,et al.A fast and elitist multiobjective genetic algorithm:NSGA-Ⅱ[J].IEEE Transactions on Evolutionary Computation,2002,6(2):182-197.
[6]Baluja S.Population-based incremental learning:A method for integrating genetic search based function optimization and competitive learning[R].Pittsburgh,Pennsylvania;Carnegie Mellon University,1994.
[7]Mishra S K,Panda G.Majhi R.A comparative performance assessment of a set of multiobjective algorithms for constrained portfolio assets selection[J].Swarm and Evolutionary Computation,2014,16:38-51.
[8]Mukhlisullina D,Passerini A,Battiti R.Learning to diversify in complex interactive multiobjective optimization[C].//Proceedings of the 10th Metaheuristics International Conference(MIC 2013),Singapore Management University,Singapore,2013.
[9]Goldberg D E,Korb B,Deb K.Messy genetic algorithms:motivation,analysis,and first results[J].Complex Systems,2010,3(3):493-530.
[10]武燕.分布估计算法研究及在动态优化问题中的应用[D].西安:西安电子科技大学,2009.
[11]周树德,孙增圻.分布估计算法综述[J].自动化学报,2007,33(2):113-124.
[12]Beasley J E.OR-Library:Distributing test problems by electronic mail[J].Journal of the Operational Research Society,1990,41(11):1069-1072.
[13]Zitzler E.Thiele L,Laumanns M,et al.Performance assessment of multiobjective optimizers:An analysis and review[J].IEEE Transactions on Evolutionary Computation,2003,7(2):117-132.
[14]Huang Renfang.Luo Xianwu.Ji Bin,et al.Multi-objective optimization of a mixed-flow pump impeller using modified NSGA-Ⅱalgorithm[J].Science China technological Sciences,2015,58(12):2122-2130.
[2]何朝林.均值—方差模型具有一般不确定性下的最优资产组合选择[J].中国管理科学,2015,23(12):63-70.
[3]苏治,秦磊,方彤.含有图结构约束的稀疏最小方差资产组合模型[J].中国管理科学,2015,23(9):65-70.
[4]Meignan D,Knust S,Frayret J M,et al.A review and taxonomy of interactive optimization methods in operations research[J].ACM Transactions on Interactive Intelligent Systems,2015,5(3):1-43.
[5]Deb K.Pratap A,Agarwal S,et al.A fast and elitist multiobjective genetic algorithm:NSGA-Ⅱ[J].IEEE Transactions on Evolutionary Computation,2002,6(2):182-197.
[6]Baluja S.Population-based incremental learning:A method for integrating genetic search based function optimization and competitive learning[R].Pittsburgh,Pennsylvania;Carnegie Mellon University,1994.
[7]Mishra S K,Panda G.Majhi R.A comparative performance assessment of a set of multiobjective algorithms for constrained portfolio assets selection[J].Swarm and Evolutionary Computation,2014,16:38-51.
[8]Mukhlisullina D,Passerini A,Battiti R.Learning to diversify in complex interactive multiobjective optimization[C].//Proceedings of the 10th Metaheuristics International Conference(MIC 2013),Singapore Management University,Singapore,2013.
[9]Goldberg D E,Korb B,Deb K.Messy genetic algorithms:motivation,analysis,and first results[J].Complex Systems,2010,3(3):493-530.
[10]武燕.分布估计算法研究及在动态优化问题中的应用[D].西安:西安电子科技大学,2009.
[11]周树德,孙增圻.分布估计算法综述[J].自动化学报,2007,33(2):113-124.
[12]Beasley J E.OR-Library:Distributing test problems by electronic mail[J].Journal of the Operational Research Society,1990,41(11):1069-1072.
[13]Zitzler E.Thiele L,Laumanns M,et al.Performance assessment of multiobjective optimizers:An analysis and review[J].IEEE Transactions on Evolutionary Computation,2003,7(2):117-132.
[14]Huang Renfang.Luo Xianwu.Ji Bin,et al.Multi-objective optimization of a mixed-flow pump impeller using modified NSGA-Ⅱalgorithm[J].Science China technological Sciences,2015,58(12):2122-2130.