Distance Weighted 2D Kernel AutoAssociation Memory Model and Its Applications
CHEN Lei1,2, WANG ChuanDong1, SUN ZhiXin1, CHEN SongCan2
1.Department of Computer Science and Technology, Nanjing University of Posts and Telecommunications, Nanjing 210003 2.Department of Computer Science and Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 210016
Abstract By using the kernel trick to modify Hopfield autoassociative memory model (HAM), a framework of kernel autoassociation memory model (KAM) is proposed. KAM makes exponential correlation associative memory (ECAM) and improved ECAM (IECAM) become two special cases. Then, the framework of distance weighted 2D kernel autoassociation memory model (DW2DKAM) is constructed by introducing distance factors to the kernels. DW2DKAM improves the storage capacity and errorcorrecting capability of KAM when recognizing binary visual images. Simulation results verify that DW2DKAM has higher storage capacity and better errorcorrecting capability than those of KAM, and outperforms the recently proposed modular HAM by Seow and Asari.
[1] Hopfield J J. Neural Network and Physical Systems with Emer
gent Collective Computational Abilities. Proc of the National Academy of Sciences of the United States of America, 1982, 79(8): 25542558 [2] Zhou Zhihua, Cao Cungen. Neural Network and Applications. Beijing, China: Tsinghua University Press, 2004: 295320 (in Chinese) (周志华,曹存根.神经网络及其应用.北京:清华大学出版社, 2004: 295320) [3] Psaltis D, Park H C. Nonlinear Discriminant Functions and Associative Memories // Proc of the AIP Conference on Neural Networks for Computing. Snowbird, USA, 1987: 370375 [4] Dembo A, Zeitouni O. General Potential Surfaces and Neural Networks. Physics Review A: General Physics, 1988, 37 (6): 21342143 [5] Chiueh T D, Goodman R M. Recurrent Correlation Associative Memories. IEEE Trans on Neural Networks, 1991, 2(2): 275284 [6] Zhang Bailing, Zhang Haihong, Ge S Z S. Face Recognition by Applying Wavelet Subband Representation and Kernel Associative Memory. IEEE Trans on Neural Networks, 2004, 15(1): 166177 [7] Nowicki D, Dekhtyarenko O. KernelBased Associative Memory // Proc of the IEEE International Joint Conference on Neural Networks. Budapest, Hungary, 2004, Ⅰ: 741744 [8] García C, Moreno J A, Sloot P M A, et al. The Kernel Hopfield Memory Network // Proc of the 6th International Conference on Cellular Automata for Research and Industry. Amsterdam, Netherlands, 2004: 755764 [9] Caputo B, Niemann H. Storage Capacity of Kernel Associative Memories // Proc of the International Conference on Artificial Neural Network. Madrid, Spain, 2002: 5156 [10] Mercer J. Functions of Positive and Negative Type, and Their Connection with the Theory of Integral Equations. Philosophical Transaction of the Royal Society of London, 1909, 209: 415466 [11] Cristianini N, ShaweTaylor J. An Introduction to Support Vector Machines and Other KernelBased Learning Methods. Cambridge, UK: Cambridge University Press, 2000 [12] Gazzaniga M S. Organization of the Human Brain. Science, 1989, 245(492): 947952 [13] Kohonen T. SelfOrganizing Maps. 2nd Edition. Berlin, Germany: SpringerVerlag, 1997 [14] Seow M J, Asari V K. Learning Using Distance Based Training Algorithm for Pattern Recognition. Pattern Recognition Letters, 2004, 25(2): 189196 [15] Boser B E, Guyon I M, Vapnik V N. A Training Algorithm for Optimal Margin Classifiers // Proc of the 5th Annual Workshop on Computational Learning Theory. San Mateo, USA, 1992: 144152
2007, Vol. 20 
