基于混沌序列遗忘神经网络算法研究管理科学与工程专业论文.docxVIP

哈尔滨工业大学管理学硕士学位论文 哈尔滨工业大学管理学硕士学位论文 -II -II- Abstract Intelligent information processing technology is a very challenging task in the scientific development. Artificial neural network is a kind of model which simulates the human neural network based on mathematical tools, which realize the simulation and approximate of biological neural networks. As an effective intelligent information processing techniques, artificial neural network has achieved remarkable development and is widely used in many fields. At the same time, the Chaos Theory, one of the pillars of the latest intelligent computing between the end of the 20th and the beginning of the 21th century, also simulates the enthusiasm of scientists of science and engineering to devote themselves to this field. With the development of computer technology, data warehouses are being widely used in enterprise information systems as a kind of subject-oriented, integrated, stable, and time varied data collection. The establishment of data warehouse enriches the historical data for decision-maker. However, it makes the decision-maker facing with the choice of trade-off in front of mass and different time span data. Forgetting is an affective way of solving this kind of problem. When comparing the model of neural network, there often exist such kind of HYPERLINK /phenomenon/ phenomenon that it affected by initial weights. The superiority of the model algorithm may be concealed or even be subverted be the sensitivity of the initial weights. Pseudo-random sequence generated by chaotic sequence can better simulate the random sequence effectively. This paper studied the basic principles of amnestic neural network, and introduce chaos theory model to it. On the basis of chaos theory we make use of logistic chaotic sequence as initial weights so as to establish an improved feed-forward neural network model Amnestic Neural Network algorithm based on Chaotic Sequence. It effectively increases the convergence speed and improves pr