2002 Fiscal Year Final Research Report Summary
Research on Fuzzy Portfolio Selection method based on Hierarchical Boltzmann Machine
| Project/Area Number |
11680459
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
社会システム工学
|
|---|
| Research Institution | Osaka Institute of Technology |
|---|
Principal Investigator |
WATADA Junzo Osaka Institute of Technology, Faculty of Engineering, Professor, 工学部, 教授 (10158610)
|
|---|
| Project Period (FY) |
1999 – 2002
|
| Keywords | Hierarchical Boltzmann machine / Meta-controlled Boltzmann machine / Fuzzy portfolio selection problem / Selection a limited number of stocks / Mean-variance analysis / Fuzzy quadratic programming / ファジィ数理計画法 |
|---|
| Research Abstract |
The research on methods of investment of stocks starts from portfolio selection model proposed by H. Markowits. But it is difficult to realize sufficiently efficient selection of a certain number of stocks out of all stocks in the market. The objective of this research is to build the improved model using the neural network which we have been studied for years.
The research results obtained in this granted research can be summarized as follows:
1) We successed to organically drive a neural network by connecting between a Boltzmann machine and a Hopfield machine. This result was presented at international Journals and international conferences.
2) In order to solve a portfolio selection problem under the consideration of the previous investing pattern, we developed the method to employ the hierarchical neural network in solving. The meta-controlled layer consisting of a Hopfield network is built in to select the appropriate number of stocks and the lower layer of a Boltzmann machine is emp
… More
loyed to effectively minimize the difference between the present investment pattern and the previous investment pattern. This model successfully obtained the solution. This result was presented at international conferences.
3) It enables us to forecast the price of a stock at the next term using Fuzzy Chaotic Short-Term Forecasting Model which the head investigator have studied. The portfolio of stocks can be solved using the forecasted price under the consideration of dealing unit at the market. This result was presented at international conferences.
4) It enables us to solve the portfolio selection problem considering a dealing unit by combining between the chaotic method and the meta-controlled Boltzmann machine. This result was presented at international conferences.
5) The above-mentioned methods are applied to solve agricultural to management, production management and project management as well as to portfolio selection problems. This result was presented at international Journals and international conferences. Less
|
