Fully Complex Multi-Layer Perceptron Network for Nonlinear Signal Processing
- 563 Downloads
Designing a neural network (NN) to process complex-valued signals is a challenging task since a complex nonlinear activation function (AF) cannot be both analytic and bounded everywhere in the complex plane ℂ. To avoid this difficulty, ‘splitting’, i.e., using a pair of real sigmoidal functions for the real and imaginary components has been the traditional approach. However, this ‘ad hoc’ compromise to avoid the unbounded nature of nonlinear complex functions results in a nowhere analytic AF that performs the error back-propagation (BP) using the split derivatives of the real and imaginary components instead of relying on well-defined fully complex derivatives. In this paper, a fully complex multi-layer perceptron (MLP) structure that yields a simplified complex-valued back-propagation (BP) algorithm is presented. The simplified BP verifies that the fully complex BP weight update formula is the complex conjugate form of real BP formula and the split complex BP is a special case of the fully complex BP. This generalization is possible by employing elementary transcendental functions (ETFs) that are almost everywhere (a.e.) bounded and analytic in ℂ. The properties of fully complex MLP are investigated and the advantage of ETFs over split complex AF is shown in numerical examples where nonlinear magnitude and phase distortions of non-constant modulus modulated signals are successfully restored.
Unable to display preview. Download preview PDF.
- 2.T. Clarke, “Generalization of Neural Network to the Complex Plane,” in Proc. of IJCNN, vol. 2, 1990, pp. 435-440.Google Scholar
- 5.D. Mandic and J. Chambers, Recurrent Neural Networks for Prediction, John Wiley and Sons, 2001.Google Scholar
- 8.N. Benvenuto, M. Marchesi, F. Piazza, and A. Uncini, “Non Linear Satellite Radio Links Equalized Using Blind Neural Networks,” in Proc. of ICASSP, vol. 3, 1991, pp. 1521-1524.Google Scholar
- 10.M. Ibnkahla and F. Castanie, “Vector Neural Networks for Digital Satellite Communications,” in Proc. of ICC, vol. 3, 1995, pp. 1865-1869.Google Scholar
- 11.A. Uncini, L. Vecci, P. Campolucci, and F. Piazza, “Complex-Valued Neural Networks with Adaptive Spline Activation Functions,” IEEE Trans. on Signal Processing, vol. 47,no. 2, 1999.Google Scholar
- 12.S. Bandito and E. Biglieri, “Nonlinear Equalization of Digital Satellite Channels,” IEEE Jour. on SAC., vol. SAC-1., 1983, pp. 57-62.Google Scholar
- 14.J. Deng, N. Sundararajan, and P. Saratchandran, “Communication Channel Equalization Using Complex-Valued Minimal Radial Basis Functions Neural Network,” in Proc. of IEEE IJCNN 2000, vol. 5, 2000, pp. 372-377.Google Scholar
- 16.S. Chen, P.M. Grant, S. McLaughlin, and B. Mulgrew, “Complex-Valued Radial Basis Function Networks,” in Proc. of third IEEE International Conference on Artificial Neural Networks,” 1993, pp. 148-152.Google Scholar
- 17.T. Kim and T. Adah, “Fully Complex Backpropagation for Constant Envelop Signal Processing,” in Proc. of IEEE Workshop on Neural Networks for Sig. Proc., Sydney, Dec. 2000, pp. 231-240.Google Scholar
- 18.T. Kim and T. Adah, “Complex Backpropagation Neural Network Using Elementary Transcendental Activation Functions,” in Proc. of IEEE ICASSP, Proc. vol. II, Salt Lake City, May 2001.Google Scholar
- 19.T. Kim and T. Adah, “Nonlinear Satellite Channel Equalization Using Fully Complex Feed-Forward Neural Networks,” in Proc. of IEEE Workshop on Nonlinear Signal and Image Processing, Baltimore, June, 2001, pp. 141-150.Google Scholar