: 对模式识别系统而言,不同的训练样本在建立模式类模型时所起的作用不同,因此必须对训练样本进行选择。而在训练样本中,边界样本的判定方式以及训练样本中包含边界样本数量的多少对分类的精度起主要作用。为此,结合基于模板匹配的脱机手写汉字识别,定义了一种通过广义置信度判定边界样本的方法,并且在此基础上建立了基于广义置信度的训练样本选择算法。通过在脱机手写汉字数据库HCL2004上进行实验,由该算法选择出的训练样本集在训练样本数减少的同时,使得系统识别率有了较大的提高,从而证实了该算法的有效性。
Abstract
In the process of training, some patterns are indispensable because they describe the characteristic of the class, but other patterns are dispensable. Sometimes, with these patterns the system performance even gets worse. So it is necessary to select the training patterns and find a more representative pattern subset. In this paper, a definition of the boundary patterns based on the generalized confidence is given, and a new algorithm of pattern selection is founded on this definition. According to the experiments on the offline handwritten Chinese character database HCL2004, the pattern subset selected by these algorithms have less patterns than the original set, but the system performance based on the subset is improved. Then the validity of the definition and these algorithms is approved.
关键词
人工智能 /
模式识别 /
广义置信度 /
样本选择 /
手写汉字识别 /
HCL2004
{{custom_keyword}} /
Key words
artificail intelligence /
pattern recognition /
generalized confidence /
pattern selection /
handwritten Chinese characters recognition /
HCL2004
{{custom_keyword}} /
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] Foody, G. M., Issue in training set selection and refinement for classification by a feedforward neural network[A]. In: IEEE International Geoscience and Remote Sensing sysmposium proceedings[C]. 1998. 1: 409-411. July.
[2] M. Wann, T. Hediger, and N. N. Greenbaun, The influence of training sets on generalization in feesforward neural networks[A]. In: Proc. International Joint Conf. On Neural Networks[C]. 1990. 3: 137-142.
[3] D Cohn, L Altlas, R Ladner, Improving generalization with active learning [J]. Machine Learning,1994,15: 201-221.
[4] A Robel, Dynamic pattern selection for faster learning and controlled generalization of neural networks [A]. ESANN[C]. 1994.
[5] Chaudhuri D., Murthy C. A., Chaudhuri B. B., Finding a subset of representative points in a dataset[J]. IEEE Trans. Syst. Man Cyber., 24, 1994, 1416-1425.
[6] Dasarathy B. V., Minimal consistent set identification for optimal nearest neighbor decision system design[J]. IEEE Trans, Syst. Man Cyber., 6, 1994, 511-516.
[7] Engelbrecht, A.P.; Cloete, I., Selective learning using sensitivity analysis[A]. In: Neural Networks Proceedings, 1998. IEEE World Congress on Computational Intelligence[C]. 1998. 2: 1150-1155.
[8] Tatiana Tambouratzis, Counter-clustering for training pattern selection[J]. The Computer Journal, 2000, 43(3): 177-190.
[9] 林晓帆,丁晓青,吴佑寿,等. 字符识别的置信度分析[J]. 清华大学学报(自然科学版),1998, 38(9): 47-50.
[10] 林晓帆,丁晓青,吴佑寿. 最近邻分类器置信度估计的理论分析[J]. 科学通报,1998, 43(3): 322-325.
[11] Gondail F, Lange E, Iwamoto T, et al, Face recognition system using local autocorrelations and multiscale integration[J]. IEEE Trans PAMI, 1996,18(10): 1024-1028.
[12] Smith S J, Handwritten character classification using nearest neighbor in large database[J]. IEEE Trans PAMI, 1994, 16(9): 915-919.
{{custom_fnGroup.title_cn}}
脚注
{{custom_fn.content}}
基金
南京师范大学211资助项目(1240702504)
{{custom_fund}}
PDF(140 KB)
