A comparative study on mirror image learning and ALSM

T. Wakabayashi1, Meng Shi1, W. Ohyama1, F. Kimura1
1Faculty of Engineering, Mie University, Tsu, Mie, Japan

Tóm tắt

In this paper, the effectiveness of a corrective learning algorithm MIL (mirror image learning) is comparatively studied with that of ALSM (average learning subspace method). Both MIL and ALSM were proposed to improve the learning effectiveness of class conditional distributions. While the ALSM modifies the basis vectors of a subspace by subtracting the autocorrelation matrix for counter classes from the one of its own class, the MIL generates a mirror image of a pattern which belongs to one of a pair of confusing classes to increases the size of the learning sample of the other class. The performance of two algorithms is evaluated on handwritten numeral recognition test for IPTP CDROMI. Experimental results show that the recognition rate of the subspace method is improved from 99.05% to 99.37% by ALSM and to 99.39% by MIL, respectively. Furthermore, the recognition rate of the projection distance method is improved from 99.13% to 99.35% by ALSM and to 99.44% by MIL.

Từ khóa

#Mirrors #Laser radar #Autocorrelation #Counting circuits #Handwriting recognition #Image generation #Testing #Euclidean distance #Vectors #Covariance matrix

Tài liệu tham khảo

10.1002/scj.4690260804 chang, 2001, LIBSVM A library for support vector machines 10.1016/B978-0-12-471150-1.50014-5 fukunaga, 1990, Introduction to statistical pattern recognition 10.1016/0031-3203(83)90064-X 10.1109/ICDAR.2001.953810 10.1007/3-540-44596-X_20 10.1109/ICPR.1998.711106 watanabe, 1973, Subspace method of pattern recognition, Proc 2nd IJCPR fukumoto, 2000, Accuracy improvement of handwritten character recognition by GLVQ, Proceedings of IWFHR'2000, 271 sato, 1996, Generalized learning vector quantization, Advances in Neural Information Processing 8 Proc of the 1995 Conference, 423 osuka, 1996, IPTP survay on handwritten numeral recognition, IPTP Research and Survey Report (English Translation) R-96-V-02 ikeda, 1983, Projection distance method for recognition of hand-written characters, Trans IPS Japan, 24, 106 oja, 1983, Subspace Method of Pattern Recognition