Synthesis of a k-winners-take-all neural network using linear programming with bounded variables

A k-winners-take-all (KWTA) problem is formulated as a linear programming (LP) problem with bounded variables. The solution set of the LP problem determines the winners. The LP problem is converted into an unconstrained optimization problem with two exact penalty functions, that is solved by using a gradient descent method implemented as a neural network. Theoretical results ensuring the convergence to the correct solution are provided.

[1]  Weiping Li,et al.  Applied Nonlinear Control , 1991 .

[2]  E. Kaszkurewicz,et al.  Matrix diagonal stability in systems and computation , 1999 .

[3]  Eugenius Kaszkurewicz,et al.  Convergence analysis of neural networks that solve linear programming problems , 2002, Proceedings of the 2002 International Joint Conference on Neural Networks. IJCNN'02 (Cat. No.02CH37290).

[4]  R. Decarlo,et al.  Variable structure control of nonlinear multivariable systems: a tutorial , 1988, Proc. IEEE.

[5]  Aleksej F. Filippov,et al.  Differential Equations with Discontinuous Righthand Sides , 1988, Mathematics and Its Applications.

[6]  Andrzej Cichocki,et al.  A new neural network for solving linear programming problems , 1996 .

[7]  Andrzej Cichocki,et al.  Neural networks for optimization and signal processing , 1993 .

[8]  B. Paden,et al.  Lyapunov stability theory of nonsmooth systems , 1993, Proceedings of 32nd IEEE Conference on Decision and Control.

[9]  Christopher Edwards,et al.  Sliding mode control : theory and applications , 1998 .

[10]  Vadim I. Utkin,et al.  Sliding Modes in Control and Optimization , 1992, Communications and Control Engineering Series.

[11]  Kiichi Urahama,et al.  K-winners-take-all circuit with O(N) complexity , 1995, IEEE Trans. Neural Networks.