Stability Of Hopfield Neural Network And Its Application In Portfolio

This paper mainly deals with three problems: make use of Radial Basis Function Neural Network to forecast the variation of stock price, stability of Hopfield Neural Network system and improving on algorithm, solving the Markowitz Model.Modern Portfolio Theory was founded by Harry·A·Markowitz in 1952, who gained Nobel price in 1990. The title of his article which was published in 1952 is "Portfolio Choice", it is a flag of Modern Portfolio Theory and which has become a hotspot of economics evolution. Artificial Neural Network Theory is a frontier field of research which develops at the worldwide scope in the middle and latter half of the 1980s, it has brought about a lot of influence on computer science, artificial intelligence and cognitive science, etc.This paper is divided into six chapters, in Chapter 1, some knowledge of Modern Portfolio Theory and Artificial Neural Network is introduced and the work of the paper is given, in Chapter 6, summing up and taking a view.In Chapter 2, introducing the progress in Modern Portfolio Theory research, we can get such a model by deriving from Markowitz Model: In Chapter 3, making use of Radial Neural Network to forecast the variation of stock price and making a comparison with the other methods of forecast, and according to the result of the forecast to get the expected yields on 100 kinds of stocks. In chapter 4, we give a detailed introduction on Hopfield Neural Network, analysis the stability of the system carefully. In Chapter 5, we mainly discuss how to solve the model (V) by using Hopfield Neural Network. We can give such energy function:The iterative equation of the interior state of Neural Network isTake the case of securities, simulate the election of it by using the improved algorithm of Hopfield Neural Network randomly. The achievement is as follows:(1)utilize Radial Neural Network to forecast the variation of stock price and compare it with other forecasting technology;(2)analysis the Hopfield Neural Network system dynamic stability comprehensively and obtain good results;(3)make use of the results what we get to improve the Hopfield Neural Network algorithm and take the case of stock selection to testify our theory;(4)solve the Markowitz Model.