Similar articles

Abstract: Neurons with complex-valued weights have stronger capability because of their multi-valued threshold logic. Neurons with such features may be suitable for solution of different kinds of problems including associative memory, image recognition and digital logical mapping. In this paper, robustness or tolerance is introduced and newly defined for this kind of neuron according to both their mathematical model and the perceptron neuron’s definition of robustness. Also, the most robust design for basic digital logics of multiple variables is proposed based on these robust neurons. Our proof procedure shows that, in robust design each weight only takes the value of i or −i, while the value of threshold is with respect to the number of variables. The results demonstrate the validity and simplicity of using robust neurons for realizing arbitrary digital logical functions.

[1] Aizenberg, I., Aizenberg, N., Butakov, C., Farberov, E., 2000. Image Recognition on the Neural Network Based on Multi-valued Neurons. Proc. 15th Int. Conf. on Pattern Recognition, p.989-992.

[2] Aizenberg, N., Aizenberg, I., 1994. CNN-like Networks Based on Multi-valued and Universal Binary Neurons: Learning and Application to Image Processing. Proc. 3rd IEEE Int. Workshop on Cellular Neural Networks and Their Applications, p.153-158.

[3] Aizenberg, N., Aizenberg, I., Krivosheev, G., 1996. Multi-valued and Universal Binary Neurons: Mathematical Model, Learning, Networks, Application to Image Processing and Pattern Recognition. Proc. 13th Int. Conf. on Pattern Recognition, p.185-189.

[4] Goh, S.L., Mandic, D.P., 2005. Nonlinear adaptive prediction of complex-valued signals by complex-valued PRNN. IEEE Trans. Signal Processing, 53(5):1827-1836.

[5] Gray, D.L., Miched, A., 1992. A training algorithm for binary feedforward neural networks. IEEE Trans. Neural Networks, 3(2):176-194.

[6] Hundiwal, A.K., 1993. An Improved Algorithm for Learning Representations in Boolean Neural Networks. Proc. 36th Midwest Symp. on Circuits and Systems, 1:592-595.

[7] Jankowski, S., Lozowski, A., Zurada, M., 1996. Complex-valued multi-state neural associative memory. IEEE Trans. Neural Networks, 7(6):1491-1496.

[8] Kalra, P.K., Mishra, D., Tyagi, K., 2007. A Novel Complex-valued Counterpropagation Network. IEEE Symp. on Computational Intelligence and Data Mining, p.81-87.

[9] Michel, H.E., Awwal, A.A.S., Rancour, D., 2006. Artificial Neural Networks Using Complex Numbers and Phase Encoded Weights—Electronic and Optical Implementations. Proc. Int. Joint Conf. on Neural Networks, p.486-491.

[10] Muezzinoglu, M.K., Guzelis, C., Zurada, J.M., 2003. A new design method for the complex-valued multistate Hopfield associative memory. IEEE Trans. Neural Networks, 14(4):891-899.

[11] Nishikawa, I., Sakakibara, K., Iritani, T., Kuroe, Y., 2005. 2 Types of Complex-valued Hopfield Networks and the Application to a Traffic Signal Control. Proc. Int. Joint Conf. on Neural Networks, p.782-787.

[12] Nishikawa, I., Iritani, T., Sakakibara, K., 2006. Improvements of the Traffic Signal Control by Complex-valued Hopfield Networks. Proc. Int. Joint Conf. on Neural Networks, p.459-464.

[13] Nishikawa, I., Hayashi, K., Sakakibara, K., 2007. Complex-valued Neuron to Describe the Dynamics after Hopf Bifurcation: an Example of CPG Model for a Biped Locomotion. Proc. Int. Joint Conf. on Neural Networks, p.327-332.

[14] Zhang, J.Y., Xu, J., 1996. Analysis and Design of Most Tolerant Logic Neural Networks. Proc. 3rd Int. Conf. on Signal Processing, 2:1425-1428.