Dissertation > Excellent graduate degree dissertation topics show

Combination of rule-based classifier

Author: ShiGuoQiang
Tutor: FanMing
School: Zhengzhou University
Course: Computer Software and Theory
Keywords: Combined Classifier Feature extraction RIPPER Principal component analysis Kappa - error Fig.
CLC: TP311.13
Type: Master's thesis
Year: 2010
Downloads: 93
Quote: 0
Read: Download Dissertation

Abstract


The classification is one of the important research topic in data mining. It is widely used in areas such as scientific experiments and business forecasting. How to improve the classification accuracy of the model is the core issue of classification, the combination of classification model than a single classification model has a distinct advantage in the theoretical and experimental. In this paper, the classification rules based depth study of the combination of rule-based classifier. Common Bagging and Boosting is mainly based on sampling with replacement. On a small sample data sets the sampling may cause loss of information, resulting in the base classifiers accurate rate, thus affecting the overall classification performance. Therefore, all data sets to establish the base classifier, to ensure the integrity of the information, so that the base classifier has a higher accuracy. Based on the above ideas, this paper presents the new method PCARules base classifier using rule-based combined classifier. Although the method using the base classifier predicted weighted voting to determine the class of the sample to be classified, but in this article to create a training data set based classifier Bagging and Boosting completely different. The method of this paper is not to be provided by sampling to create the data set for the group classifier, but random characteristic is divided into k subsets use PCA to obtain a main component of each subset to form a new feature space, and all the training data is mapped to a new feature space as a base classifier training set. On 30 randomly selected data set of the UCI Machine Learning Repository: Experiments show that the algorithm can not only significantly improve the classification performance of the rule-based classification method, and compared with the traditional combination of Bagging and Boosting method, the algorithm in Most of the data sets have a higher classification accuracy. This paper studies the impact of the difference between its base classifier accuracy the PCARules model performance. Observed three randomly selected set of data on the experimental results, we found that: the differences between the base classifiers does not guarantee the accuracy of combined classifiers (AdaBoost) In contrast, moderate differences and strong the complementarity can often produce the better combination classifier (PCARules); same time, the base classifier accuracy of the performance of the combined classifiers may also have a great impact example, in the PCARules, the base classifier accuracy significantly higher than Bagging, AdaBoost method in the base classifier.

Related Dissertations

  1. Research on Automatic Detection Algorithm for Substructure Distress of Highway Pavement Based on SVM,U418.6
  2. ISAR Imaging Simulation of Space Targets and Target Recognition Based on ISAR Images,TN957.52
  3. Research on Feature Extraction and Classification of Pulse Waveform for Cholecystitis and Nephrotic Syndrome Diagnosis,TP391.41
  4. Application of Q-Learning in the Content-Based Image Retrieval Technology,TP391.41
  5. Research on Transductive Support Vector Machine and Its Application in Image Retrieval,TP391.41
  6. Research on Feature Extraction and Classification of Tongue Shape and Tooth-Marked Tongue in TCM Tongue Diagnosis,TP391.41
  7. Research on Visual Measurement for Spacecraft Rendezvous and Approach,TP391.41
  8. Research on the Image Real-Time Acquisition, Storage and Image Processing System,TP391.41
  9. Feature Extraction, Selection and Combination in Lipreading,TP391.41
  10. Multi-currency Notes Technology Research and Implementation,TP391.41
  11. The Research on Paper Currency Classification Method Based on Harr-Like Feature and Minimal Ball Including Samples,TP391.41
  12. Pavement Distress Recognition Based on Image,TP391.41
  13. Research on Visual Detection and Tracking of Mobile Robots,TP242.62
  14. Research on Fusion Algorithm of Hyper Spectral and High Spatial Resolution Remote Sensing Image,TP751
  15. Application of Improved Principal Component Analysis Algorithm in Course Construction,G642.4
  16. An Approach for Identifying a Plant Resistance Gene Based on the Random Forest,Q943
  17. Tobacco Diseases Auto-Recognition Research Based on Image Processing Technology,S435.72
  18. Research on Nondestructive Detection Technology for External Qualities of Papayas Based-on Vision,S667.9
  19. Research on Identification System of Cashmere and Wool Fiber,TS101.921
  20. Research for Infrared Image Target Identification and Tracking Technology,TP391.41
  21. The Compression and Fusion Technique Research of Underwater Target Feature,TN911.7

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer software > Program design,software engineering > Programming > Database theory and systems
© 2012 www.DissertationTopic.Net  Mobile