Dissertation > Excellent graduate degree dissertation topics show

Prediction of Financial Time-Series Based on RBF Neural Network

Author: ZhuYiJia
Tutor: XuNanShan
School: Beijing University of Chemical Technology
Course: Applied Computer Technology
Keywords: RBF Nearest neighbor clustering Overfitting When the sequence
CLC: TP183
Type: Master's thesis
Year: 2010
Downloads: 248
Quote: 5
Read: Download Dissertation

Abstract


In the past few years, various researchers and financial analysts stressed the nonlinear analysis in financial market activities in significance. Taking into account the financial analysis of a new approach seems to be necessary - Nonlinear analysis of financial markets integration. Neural network computing developers a new work based on mathematical finance nonlinear analysis method, because the method of its own good learning, error correction capability and nonlinear approximation ability, in the time series prediction play an important effect. Time-series studies predict that one of the applications, when the financial time series data is an important part of the sequence, and is closely related with people's lives. Financial Time Series Forecasting can research on people's risk investment, state regulation of the economy plays a certain role in the reference. This paper describes the most commonly used time series forecasting model, and then describe the various neural network model, and highlights the RBF neural network for financial time series prediction model, while noting that the defects RBF neural networks, and on this basis put forward an improved RBF neural network model. Improved RBF model is mainly in the center and on the selection of radial basis width has been adjusted, the original nearest neighbor clustering to first enter the clustering adjust the input vector to the class as a central point for all vector of mean vector vector as the center of the cluster, based on the radial width of the selection is selected using iterative optimization method to choose. The improved RBF neural network used in financial time series prediction, and by comparing the original RBF neural network using the predictions made and found improved network in the prediction accuracy and training efficiency are improved. Verify the feasibility of the algorithm Finally, the improved algorithm may exist overfitting adjusted to eliminate redundancy in the center of the hidden layer, reduce the complexity of network structure and enhance the performance of the unknown model predictions. Each time you add a dynamic center and the center has been some comparison to determine whether redundancy. When both the center of the vector inner product approximation at 1:00, then remove one of the dynamic center of this input vector attributed closest distance clustering. Through experiments, we found that by eliminating over-fitting, removing redundant can be a good center.

Related Dissertations

  1. Analysis of Multiple Access Interference in Asynchronous Direct Sequence Spread Spectrum Communication System Using Gold Sequence,TN914.53
  2. Research and Hardware Simulation of Multiuser Spread Spectrum Code Acquisition Based on FPGA,TN914.42
  3. Design of Weapon Detection Device Control System,TP183
  4. Research on the Constructions of Several Classes of Sequence Based on Interleaving Techniques,TN911
  5. Study of Synchronous Generator Excitation Control Based on Neural Network Identification,TM31
  6. Survey in Nanjing and Analysis of the Mitochondrial Genome of Frankliniella Occidentalis (Pergande),S433
  7. Gene Prediction and Sequence Analysis of Insect OBP CSP and sid-1,Q78
  8. Genetic Variation Analysis of Porcine Reproductive and Respiratory Syndrome and Eukaryotic Expression of Pig Interferon α,S858.28
  9. Isolation and Identification of Infectious Bronchitis Virus and Sequence Analysis of Its S1 Gene and N Gene,S852.65
  10. Cloning 1Dx5 Gene from the Xinjiang Wheat Variety and Constructing Its Expression Vector,S512.1
  11. Mitochondrial Genome Sequence Analysis of a Brassica Juncea Landrace,S565.4
  12. Molecular Epidemiology about White Spot Syndrome in Jiangsu Province,S945.1
  13. Identification of Newcastle Disease Virus from Chicken and the Genome Molecular Characeristics of HN09-68 and HN09-83 Strains,S852.65
  14. The Identification of Fault Type in Transmission Lines Based on Neural Network,TP183
  15. The Formation Mechanism and Distribution Regularity of Reef-Bank Reserviors of Lower Palaeozoic in North China,P618.13
  16. Translocation Chromosome 1BL.1RS and Stripe Rust Resistant Varieties of Differential Gene Expression Analysis and the Study of Change in Physiological Parameters,S512.1
  17. Comparison of Chinese and Korean Word Order Its Chinese Reading Instruction Program,H195
  18. An Analysis of Influential Factors on Gold Price Change,F832.54;F724.5
  19. Research on Colorization Algorithm for Night Vision Video Sequence,TP391.41
  20. Research on Modeling Technique of Speech Signal Based on Compressed Sensing,TN912.3
  21. Analysis of Rotor Temperature Field of Hydro Generator on Asymmetric Operation Condition,TM312

CLC: > Industrial Technology > Automation technology,computer technology > Automated basic theory > Artificial intelligence theory > Artificial Neural Networks and Computing
© 2012 www.DissertationTopic.Net  Mobile