Generalized Lyapunov approach for convergence of neural networks with discontinuous or non-Lipschitz activations. (English) Zbl 1103.34044

The authors consider a class of discontinuous neural networks described by a system of diffential equations of the form \[ \dot x=Bx+Tg(x)+I, \] where \(x\in{\mathbb R}^n\) is the vector of neuron state variables, \(B\) is a diagonal matrix with negative coefficients modeling the neuron selfinhibitions, \(T\) is the matrix of neuron interconnections, \(g(x)\) is the neuron activation function and \(I\in{\mathbb R}^n\) is the vector of neuron biasing inputs. The main hypothesis is that the components of \(g(x)\) are bounded, non-decreasing and piece-wise continuous functions. Then, a solution must be understood in the sense of Filippov. In this context, new results on global exponential convergence and global convergence in finite time are proved. The proofs make use of a generalized Lyapunov-like approach that could be of independent interest for proving convergence of other nonsmooth dynamical systems.


34D23 Global stability of solutions to ordinary differential equations
92B20 Neural networks for/in biological studies, artificial life and related topics
Full Text: DOI


[1] Forti, M.; Nistri, P., Global convergence of neural networks with discontinuous neuron activations, IEEE trans. circuits syst. I, 50, 11, 1421-1435, (2003) · Zbl 1368.34024
[2] Hirsch, M., Convergent activation dynamics in continuous time networks, Neural netw., 2, 331-349, (1989)
[3] Marcus, C.M.; Westervelt, R.M., Stability of analog neural network with delay, Phys. rev. A, 39, 347-359, (1989)
[4] Kaszkurewicz, E.; Bhaya, A., On a class of globally stable neural circuits, IEEE trans. circuits syst. I, 41, 171-174, (1994) · Zbl 0843.93063
[5] Forti, M.; Tesi, A., New conditions for global stability of neural networks with application to linear and quadratic programming problems, IEEE trans. circuits syst. I, 42, 7, 354-366, (1995) · Zbl 0849.68105
[6] Liang, X.-B.; Wang, J., An additive diagonal stability condition for absolute stability of a general class of neural networks, IEEE trans. circuits syst. I, 48, 11, 1308-1317, (2001) · Zbl 1098.62557
[7] Liang, X.-B.; Si, J., Global exponential stability of neural networks with globally Lipschitz continuous activations and its application to linear variational inequality problem, IEEE trans. neural netw., 12, 2, 349-359, (2001)
[8] Arik, S., Global robust stability of delayed neural networks, IEEE trans. circuits syst. I, 50, 1, 156-160, (2003) · Zbl 1368.93490
[9] Hu, S.; Wang, J., Absolute exponential stability of a class of continuous-time recurrent neural networks, IEEE trans. neural netw., 14, 1, 35-45, (2003)
[10] Cao, J.; Wang, J., Absolute exponential stability of recurrent neural networks with Lipschitz-continuous activation functions and time delays, Neural netw., 17, 379-390, (2004) · Zbl 1074.68049
[11] Cao, J.; Lianga, J.; Lamb, J., Exponential stability of high-order bidirectional associative memory neural networks with time delays, Invent. math., 136, 75-87, (1999)
[12] Qi, H.; Qi, L., Deriving sufficient conditions for global asymptotic stability of delayed neural networks via nonsmooth analysis, IEEE trans. neural netw., 15, 99-109, (2004)
[13] Yucel, E.; Arik, S., New exponential stability results for delayed neural networks with time varying delays, Physica D, 191, 314-322, (2004) · Zbl 1070.34099
[14] Liao, X.; Li, C., An LMI approach to asymptotical stability of multi-delayed neural networks, Physica D, 200, 139-155, (2005) · Zbl 1078.34052
[15] Żak, S.H.; Upatising, V.; Hui, S., Solving linear programming problems with neural networks: A comparative study, IEEE trans. neural netw., 6, 94-104, (1995)
[16] Forti, M.; Nistri, P.; Quincampoix, M., Generalized neural network for nonsmooth nonlinear programming problems, IEEE trans. circuits syst. I, 51, 9, 1741-1754, (2004) · Zbl 1374.90356
[17] Bertsekas, D.P., Necessary and sufficient conditions for a penalty method to be exact, Math. program., 9, 87-99, (1975) · Zbl 0325.90055
[18] Chong, E.K.P.; Hui, S.; Żak, S.H., An analysis of a class of neural networks for solving linear programming problems, IEEE trans. automat. control, 44, 1095-2006, (1999)
[19] Ferreira, L.V.; Kaszkurewicz, E.; Bhaya, A., Solving systems of linear equations via gradient systems with discontinuous right hand sides: application to LS-SVM, IEEE trans. neural netw., 16, 2, 501-505, (2005)
[20] Gavaldá, R.; Siegelmann, H.T., Discontinuities in recurrent neural networks, Neural comput., 11, 715-745, (1999)
[21] Siegelmann, H.T.; Sontag, E.D., Analog computation via neural networks, Theoret. comput. sci., 131, 331-360, (1994) · Zbl 0822.68029
[22] Michel, A.N.; Miller, R.K., Qualitative analysis of large scale dynamical systems, (1977), Academic New York · Zbl 0358.93028
[23] Michel, A.N.; Farrell, J.A.; Porod, W., Qualitative analysis of neural networks, IEEE trans. circuits syst., 36, 2, 229-243, (1989) · Zbl 0672.94015
[24] Hershkowitz, D., Recent directions in matrix stability, Linear algebr. appl., 171, 161-186, (1992) · Zbl 0759.15010
[25] Berman, A.; Plemmons, R.J., Nonnegative matrices in the mathematical sciences, (1979), Academic New York · Zbl 0484.15016
[26] Clarke, F.H., Optimization and non-smooth analysis, (1983), John Wiley & Sons New York
[27] Filippov, A.F., Differential equations with discontinuous right-hand side, Transl. American math. soc., 42, 199-231, (1964) · Zbl 0148.33002
[28] Liang, X.-B.; Wang, J., Absolute exponential stability of neural networks with a general class of activation functions, IEEE trans. circuits syst. I, 47, 8, 1258-1263, (2000) · Zbl 1079.68592
[29] Paden, B.E.; Sastry, S.S., Calculus for computing filippov’s differential inclusion with application to the variable structure control of robot manipulator, IEEE trans. circuits syst., 34, 73-82, (1987) · Zbl 0632.34005
[30] Utkin, V.I., Sliding modes and their application in variable structure systems, (1978), MIR Publishers Moscow · Zbl 0398.93003
[31] Aubin, J.P.; Cellina, A., Differential inclusions, (1984), Springer-Verlag Berlin
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. It attempts to reflect the references listed in the original paper as accurately as possible without claiming the completeness or perfect precision of the matching.