Abstract
Hebbian-type associative memory is characterized by its simple architecture. However, the hardware implementation of Hebbian-type associative memories is normally complicated when there are a huge number of patterns stored. To simplify the interconnection values of a network, a nonlinear quantization strategy is presented. The strategy takes into account the property that the interconnection values are Gaussian distributed, and divides the interconnection weight values into a small number of unequal ranges accordingly. Interconnection weight values in each range contain information equally and each range is quantized to a value.
The equation of probability of direct convergence was derived. The probability of direct convergence of nonlinear quantized networks with a small number of ranges is compatible with their original networks. The effects of linear and nonlinear quantization were also assessed in terms of recall capability, information capacity, and number of bits storing interconnection values saved by quantization. The performance of the proposed nonlinear quantization strategy is better than that of the linear quantization while retaining a recall capability that is compatible with its original network. The proposed approach reduces the number of connection weights and the size of the chip areas of a Hebbian-type associative memory while approximately retaining its performance.
Similar content being viewed by others
References
Verleysen M, Sirletti B (1989) A high-storage capacity content-addressable memory and its learning algorithm. IEEE Trans Circuits Syst 36(5):762–766
Kamio T, Fujisaka H, Morisue M (2001) Backpropagation algorithm for LOGic oriented neural networks with quantized weights and multilevel threshold neurons. IEICE Trans Fundam E84-A:705–712
Heittmann A, Ruckert U (2002) Mixed mode VLSI implementation of a neural associative memory. Analog Integr Circuits Signal Process 30(2):159–172
Kamio T, Fujisaka H, Morisue M (2002) Associative memories using interaction between multilayer perceptrons and sparsely interconnected neural networks. IEICE Trans Fundam E85-A:1220–1228
Kamio T, Morisue M (2003) A synthesis procedure for associative memories using cellular neural networks with space-invariant cloning template library. In: Proceedings of the international joint conference on neural networks, pp 885–890
Ho C, Ling B, Lam HK, Nasir M (2008) Global convergence and limit cycle behavior of weights of perceptron. IEEE Trans Neural Netw 19(6):938–947
Hopfield JJ, Tank DW (1985) Neural computation of decisions in optimization problems. Biol Cybern 52(3):141–152
Tank DW, Hopfield JJ (1986) Simple optimization networks: A/D converter and a linear programming circuit. IEEE Trans Circuits Syst CAS-33:533–541
Sussner P, ME Valle (2006) Gray-scale morphological associative memories. IEEE Trans Neural Netw 17(3):559–570
Zhang H, Huang W, Huang Z, Zhang B (2005) A kernel autoassociator approach to pattern classification. IEEE Trans Syst Man Cybern, Part B, Cybern 35(3):593–606
Sussner P, Valle ME (2007) Morphological and certain fuzzy morphological associative memories for classification and prediction. In: Kaburlasos V, Ritter G (eds) Computational intelligence based on lattice theory. Springer, Heidelberg, pp 149–172
Costantini G, Casali D, Perfetti R (2003) Neural associative memory storing gray-coded gray-scale images. IEEE Trans Neural Netw 14(3):703–707
Costantini G, Casali D, Perfetti R (2006) Associative memory design for 256 gray-level images using a multilayer neural network. IEEE Trans Neural Netw 17(2):519–522
Lee DL (2006) Improvement of complex-valued Hopfield associative memory by using generalized projection rules. IEEE Trans Neural Netw 17(5):1341–1347
Monteros Rave de los A, Azuela JHS (2008) A bidirectional hetero-associative memory for true-color patterns. Neural Process Lett 28(3):131–153
Valle ME (2009) A class of sparsely connected autoassociative morphological memories for large color images. IEEE Trans Neural Netw 20(6):1045–1050
Zheng P, Zhang J, Tang W (2010) Color image associative memory on a class of Cohen-Grossberg networks. Pattern Recognit 43(10):3255–3260
Zhang BL, Zhang H, Ge SS (2004) Face recognition by applying wavelet subband representation and kernel associative memory. IEEE Trans Neural Netw 15(1):166–177
Zhang H, Zhang B, Huang W, Tian Q (2005) Gabor wavelet associative memory for face recognition. IEEE Trans Neural Netw 16(1):275–278
Zhang D, Zuo W (2007) Computational intelligence-based biometric technologies. IEEE Comput Intell Mag 2(2):26–36
Han SJ, Oh SY (2009) A new paradigm for real-time parallel storage and recognition of patterns based on a hierarchical organization of associative memories utilizing Walsh function encoding. Appl Intell 31(3):305–317
Amit DJ (1989) Modeling brain function: the world of attractor neural networks. Cambridge University Press, Cambridge
Sompolinsky H (1987) The theory of neural networks: The Hebb rule and beyond. Heidelberg Colloquium on Glassy Dyn 275:485–527
Chung PC, Krile TF (1992) Characteristics of Hebbian-type associative memories having faulty interconnections. IEEE Trans Neural Netw 3(6):969–980
Chung PC, Krile TF (1995) Reliability characteristics of quadratic Hebbian-type associative memories in optical and electronic network implementations. IEEE Trans Neural Netw 6(2):357–367
Guzmán E, Alvarado S, Pogrebnyak O, Sánchez LP, Yánez C (2007) Hardware implementation of image recognition system based on morphological associative memories and discrete wavelet transform. LNCS, vol 4827. Springer, Berlin, pp 664–677
Guzmán E, Alvarado S, Pogrebnyak O, Yánez C (2007) Image recognition processor based on morphological associative memories. In: IEEE proceedings of the electronics, robotics and automotive mechanics conference, pp 260–265
Ahmadi A, Mattausch HJ, Abedin MA, Saeidi M, Koide T (2011) An associative memory-based learning model with an efficient hardware implementation in FPGA. Expert Syst Appl 38:3499–3513
Szabó T, Horváth G (2004) An efficient hardware implementation of feed-forward neural networks. Appl Intell 21(2):143–158
Jiang M, Gielen G (2008) Analysis of quantization effects on high-order function neural networks. Appl Intell 28(1):51–67
Wang JH (1998) Principal interconnections in higher order Hebbian-type associative memories. IEEE Trans Knowl Data Eng 10(2):342–344
Chung PC, Tsai CT, Sun YN (1994) Characteristics of Hebbian-type associative memories with quantized interconnections. IEEE Trans Circuits Syst 41(2):168–171
Davey N, Frank R, Hunt S, Adams R, Calcraft L (2004) High capacity associative memory models—binary and bipolar representation. In: Proceedings of the eighth IASTED international conference artificial intelligence and soft computing, Marbella, Spain, pp 392–397
Chung PC, Tsai CT, Sun YN (1994) Linear quantization of Hebbian-type associative memories in interconnection implementation. In: IEEE international conference on neural networks, Orlando, pp 1092–1097
McEliece RJ, Posner EC (1987) The capacity of the Hopfield associative memory. IEEE Trans Inf Theory 33(4):461–482
Willshaw DJ, Buneman OP, Longuet-Higgins HC (1969) Nonholographic associative memory. Nature 222:960–962
Palm G, Sommer FT (1992) Information capacity in recurrent McCulloch–Pitts networks with sparsely coded memory states. Network 3(2):177–186
Wang JH, Krile TF, Walkup JF (1990) Determination of Hopfield associative memory characteristics using a single parameter. Neural Netw 3(3):319–331
Walkup JF, Krile TF (1991) Programmable optical quadratic neural networks. Technical Report, Department of Electrical Engineering, Texas Tech University
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Liaw, C., Tsai, CT. & Ko, CH. Nonlinear quantization on Hebbian-type associative memories. Appl Intell 36, 824–833 (2012). https://doi.org/10.1007/s10489-011-0299-7
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10489-011-0299-7