Abstract A method for rule extraction from neural networks based on the functional point of view is studied. The key algorithms are introduced, including the sort and selection of features, the discretization of continuous attributes, the generation of training samples of neural network (NN), the training of NN, the generation of the instance samples from the trained NN, and the rule extraction. The UCI data and the population classifying data are used to verify the rule extraction method. The results show the correction and effectivity of the proposed method.
[1] Yu Heji, Chen Changzhen, Zhang Sheng, et al. Intelligent Diagnosis Based on Neural Network. Beijing, China: Metallurgical Industry Press, 2000 (in Chinese) (虞和济,陈长征,张 省,等.基于神经网络的智能诊断.北京:冶金工业出版社, 2000) [2] Wu Jinpei, Xiao Jianhua. Intelligent Fault Diagnosis and Expert System. Beijing, China: Science Press, 1997 (in Chinese) (吴今培,肖建华.智能故障诊断与专家系统.北京:科学出版社, 1997) [3] Gallant S I. Connectionist Expert Systems. Communications of the ACM, 1988, 31(2): 152-169 [4] Saito K, Nakano R. Medical Diagnostic Expert System Based on PDP Model // Proc of the IEEE International Conference on Neural Networks. San Diego, USA, 1988, Ⅰ: 255-262 [5] Fu L M. Rule Learning by Searching on Adapt Nets // Proc of the 9th National Conference on Artificial Intelligence. Anaheim, USA, 1991: 590-595 [6] Zhou Zhihua, He Jiazhou, Yin Xuri, et al. A Statistics-Based Approach for Rule Extraction from Neural Networks. Journal of Software, 2001, 12(2): 263-269 (in Chinese) (周志华,何佳洲,尹旭日,等.一种基于统计的神经网络规则抽取方法.软件学报, 2001, 12(2): 263-269) [7] Kantardzic M. Data Mining Concepts, Models, Methods, and Algorithms. New York, USA: IEEE Press, 2002 [8] Pawlak Z. Rough Set. International Journal of Information and Computer Science, 1982, 11(5): 341-356 [9] Wang Guoyin.Rough Set Theory and Knowledge Acquisition. Xi'an, China, Xi'an Jiaotong University Press, 2001 (in Chinese) (王国胤.Rough集理论与知识获取.西安:西安交通大学出版社, 2001) [10] Ryszard N, Roman S, Jerzy S. Evaluation of Vibroacoustic Diagnostic Symptoms by Means of the Rough Sets Theory. Computers in Industry, 1992, 20(2): 141-152 [11] Nguyen H S, Skowron A. Quantization of Real Values Attributes, Rough Set and Boolean Reasoning Approaches // Proc of the 2nd Joint Annual Conference on Information Science. Wrightsville Beach, USA, 1995: 34-37 [12] Chen Guo. Structure Self-Adaptive Neural Network Model Realizing Structural Risk Minimization Principle. Chinese Journal of Scientific Instrument, 2007, 28(10): 1874-1879 (in Chinese) (陈 果.一种实现结构风险最小化思想的结构自适应神经网络模型.仪器仪表学报, 2007, 28(10): 1874-1879) [13] Zhou Zihua, Jiang Yuan, Chen Shifu. Extracting Symbolic Rules from Trained Neural Network Ensembles. AI Communications, 2003, 16(1): 3-15 [14] Chang Liyun, Wang Guoyin, Wu Yu. An Approach for Attribute Reduction and Rule Generation Based on Rough Set Theory. Journal of Software, 1999, 10(11): 1207-1211 (in Chinese) (常犁云,王国胤,吴 渝.一种基于Rough Set理论的属性约简及规则提取方法.软件学报, 1999, 10(11): 1207-1211) [15] Wu Fubao, Li Qi, Song Wenzhong. Inductive Learning Approach to Knowledge Representation System Based on Rough Set Theory. Control and Decision, 1999, 14(3): 206-211 (in Chinese) (吴福保,李 奇,宋文忠.基于粗集理论知识表达系统的一种归纳学习方法.控制与决策, 1999, 14(3): 206-211) [16] Chen Zhaoqian, Liu Hong, Zhou Rong, et al. A Hybrid Algorithm for Multi-Concept Acquisition and Its Application. Chinese Journal of Computers, 1996, 19(10): 753-761 (in Chinese) (陈兆乾,刘 宏,周 戎,等.一种混合型多概念获取算法HMCAP及其应用.计算机学报, 1996, 19(10): 753-761)
2008, Vol. 21 
