Qi ya-Li
School of Electrical Engineering and Automation, Tianjin University, Tianjin 300072, China
Li ye-Li
The Computer Science and Technology Department, Beijing Institute of Graphic Communication, Beijing102600, China
ABSTRACT
For the characteristics of printing malfunction diagnose system, a model to classify printing fault based on Incremental Reduced Support Vector Machine (IRSVM) and C4.5 is discussed. IRSVM is an improved method based on Support Vector Machine (SVM) which has been promising method to classify for its solid mathematical foundation. However it is not favored for large-scale, because the training complexity of SVM is highly dependent on the size of data set. This study uses IRSVM to classify root-classes, then uses C4.5 algorithm for further diagnosis to remedy the defect of IRSVM in classing subclasses. The hybrid method makes fully use of the IRSVM efficiency in multidimensional character space but it also brings the accuracy of C4.5 algorithm into full play. That is suited to class the complicated print faults. Computational results indicate the hybrid method has a good efficiency for adjustable printing fault and its computational times as well as its memory usage are much smaller than those of conventional SVM.
PDF References Citation
How to cite this article
Qi ya-Li and Li ye-Li, 2013. Printing Fault Classification Based on Hybrid Method. Information Technology Journal, 12: 5771-5774.
DOI: 10.3923/itj.2013.5771.5774
URL: https://scialert.net/abstract/?doi=itj.2013.5771.5774
DOI: 10.3923/itj.2013.5771.5774
URL: https://scialert.net/abstract/?doi=itj.2013.5771.5774
REFERENCES
- Sabzekar, M. and M. Naghibzadeh, 2013. Fuzzy c-means improvement using relaxed constraints support vector machines. Applied Soft Comput., 13: 881-890.
CrossRefDirect Link - Wu, M., B. Scholkopf and G. Bakir, 2006. A direct method for building sparse kernel learning. J. Mach. Learn. Res., 7: 603-624.
Direct Link - Lee, Y.J. and O.L. Mangasarian, 2001. SSVM: A smooth support vector machine. Computational optimization and applications. Comput. Optim. Applic., 20: 5-22.
CrossRefDirect Link