Dissertation > Excellent graduate degree dissertation topics show

A sample based on clustering neural network self- learning system

Author: HeQing
Tutor: HuYingSong
School: Huazhong University of Science and Technology
Course: Applied Computer Technology
Keywords: Neural Network Clustering algorithm Samples from the study Generalization capability
CLC: TP391.6
Type: Master's thesis
Year: 2011
Downloads: 25
Quote: 0
Read: Download Dissertation

Abstract


Neural network generalization ability is an important aspect affecting their availability , how to improve the generalization ability of growing concern. A common situation is that some of the neural network in the training phase to obtain good training accuracy, but in practical application stage has produced a large error , which largely affects the availability of the network . Because in practice , the focus of concern is not the system input and output samples of known fitting ability, but known to the system input and output of the degree of reflection . Analysis and comparison of the current commonly used to improve neural network generalization of several methods , the lack of these methods , to improve the training sample set from the completeness of departure , presents a neural network based clustering sample self-learning system . The clustering algorithm ideas introduced into the system , the actual use of the neural network during the actual input data for analysis. Guarantee does not affect the system in real-time forecasting of the premise, through clustering algorithm to get the current conditions representative data points. Analysis of these data points derived representative sample points may be new , be related processing , constitute a new sample added to the sample concentration , re- train the network . By this method makes the network more suitable for the actual situation in the current environment , improve the prediction accuracy . In Visual Studio 2005 development platform based on clustering neural network self-learning system for testing samples . To fire risk monitoring as a test case , and did not use samples from learning neural network prediction method comparison, from the prediction accuracy for performance testing. Test results show that the use of neural networks based on clustering samples of self-learning program in the prediction of dangerous conditions , the timely analysis of the actual situation , thus improving network generalization ability , reducing forecast error .

Related Dissertations

  1. Development of the Platform for Compressor Optimization Design and Aerodynamic Optimization Design in the Transonic Compressor,TH45
  2. Research on Feature Extraction and Classification of Tongue Shape and Tooth-Marked Tongue in TCM Tongue Diagnosis,TP391.41
  3. Study on Virtual Detector of Infrared Hyper-Spectral Image,TP391.41
  4. The Application of Fuzzy Control and Neural Network in Planar Double Inverted Pendulum,TP273.4
  5. Research of Visual Servoing in 4-DOF Robotic Manipulator,TP242.6
  6. Research on Visual Servo System of Mechanical ARM,TP242.6
  7. Design of Weapon Detection Device Control System,TP183
  8. Municipal tourism land use planning environmental impact assessment,X820.3
  9. Research on the Inteligent System of High Performance Concrete Mix Design in Zhujiang Triangle Area,TU528
  10. Study on Taste Characteristic of Taste Peptide Enzymatic Production from Oyster Base on A Neural Network Method,TS254.4
  11. Smart Control Research in Paddy Drying Based on BP Neural Network,S226.6
  12. BP network optimization based on genetic algorithm optimization of the biodiesel process,TE667
  13. Fire Fighting System Research for the Offshore Platform,U698.4
  14. The Research on Evaluation of Living Status Systems of Expressway Relocated People,D523
  15. The Application of Ant Colony Algorithm in Meteorological Satellite Cloud Pictures Segmentation,TP391.41
  16. Research on Clustering Algorithm Based on Mutation Particle Swarm Optimization,TP18
  17. Research on K-means Optimization Clustering Algorithm,TP311.13
  18. Research on Fuzzy C-Mean Clustering Algorithm Based on Particle Swarm Optimization and Shuffled Frog Leaping Algorithm,TP18
  19. Research on Clustering Algorithm Based on Genetic Algorithm and Rough Set Theory,TP18
  20. Research on Inspection Technology of Dehydrated Garlic Slice Based on Computer Vision,TP391.41
  21. Research and Application of Safety Assessment Method Based on Neural Network,X937

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Teaching machine, learning machine
© 2012 www.DissertationTopic.Net  Mobile