研究了神经网络的输出变量对输入变量和权重变量的敏感性定义,通过分析这些敏感性定义的本质,把神经网络的敏感性定义分为3类:分析学,几何学,物理学.在这3种敏感性定义分类中,对分析学和物理学定义的敏感性进行了模拟实验,实验结果表明,物理学定义的敏感性更适合应用在径向基函数神经网络删除冗余属性上.在解决其他网络的冗余属性问题上,需要进一步研究.
[1] 李士勇. 模糊控制、神经控制和智能控制论 [M]. 哈尔滨:哈尔滨工业大学出版社 1998.
[2] HASHEM S. Sensitivity analysis for feedforward artificial neural networks with differentiable activation functions [A]. Baltimore 1992.
[3] ZURADA J M, MALINOWSKI A, USUI S. Perturbation method for deleting redundant inputs of perceptron networks [J]. Elsevier Science B V 1997, 14(02).
[4] WANG Xizhao, LI Chunguo. A new definition of sensitivity for RBFNN and Its applications to feature reduction [J]. Lecture Notes in Computer Science 2005, 3496.
[5] Jin Young Choi, Chong-Ho Choi. Sensitivity analysis of multilayer perceptron with differentiable activation functions [J]. IEEE Transactions on Neural Networks 1992, 1(1).
[6] YEUNG D S, SUN Xuequan. Using function approximation to analyze the sensitivity of MLP with antisymmetric squashing activation function [J]. IEEE Transactions on Neural Networks 2002, 13(01).
[7] Karayiannis Nicolaos B.. Reformulated radial basis neural networks trained by gradient descent [J]. IEEE Transactions on Neural Networks 1999, 3(3).
[8] Stevenson M., Winter R.. Sensitivity of feedforward neural networks to weight errors [J]. IEEE Transactions on Neural Networks 1990, 1(1).
[9] Cheng A.Y., Yeung D.S.. Sensitivity analysis of neocognitron [J]. IEEE transactions on systems, man and cybernetics, Part C. Applications and reviews: A publication of the IEEE Systems, Man, and Cybernetics Society 1999, 2(2).
[10] Piche S.W.. The selection of weight accuracies for Madalines [J]. IEEE Transactions on Neural Networks 1995, 2(2).
[11] NG W W Y, YEUNG D S, RAN Q. Statistical output sensitivity to input and weight perturbations of radial basis function neuron networks [A]. Twnisia 2002.
[12] ZENG Xiaoqin, YEUNG D S. Sensitivity analysis of multilayer perceptron to input and weight perturbations [J]. IEEE Transactions on Neural Networks 2001, 12(06).
[13] Xiaoqin Zeng, Daniel S. Yeung. A Quantified Sensitivity Measure for Multilayer Perceptron to Input Perturbation [J]. Neural computation 2003, 1(1).
[14] Shi D, Yeung DS, Gao J. Sensitivity analysis applied to the construction of radial basis function networks. [J]. Neural Networks: The Official Journal of the International Neural Network Society 2005, 7(7).
[15] NG W W Y, YEUNG D S. Input dimensionality reduction for Radial Basis Neural Network classification problems using sensitivity analysis [A]. 北京 2002.
[16] NG W W Y, CHANG R K C, YEUNG D S. Dimensionality reduction for denial of service detection problems using RBFNN output sensitivity [A]. 陕西西安 2003.
