论文信息 - Observability of Neural Network Behavior

Observability of Neural Network Behavior

We prove that except possibly for small exceptional sets, discrete-time analog neural nets are globally observable, i.e. all their corrupted pseudo-orbits on computer simulations actually reflect the true dynamical behavior of the network. Locally finite discrete (boolean) neural networks are observable without exception.

Max H. Garzon | Fernanda Botelho | M. Garzon | F. Botelho

[1] Max H. Garzon,et al. Global Dynamics in Neural Networks II , 1990, Complex Syst..

[2] J. Yorke,et al. Pseudo-orbit shadowing in the family of tent maps , 1988 .

[3] Eduardo D. Sontag,et al. Identi ability of discrete-time neural networks , 1993 .

[4] Héctor J. Sussmann,et al. Uniqueness of the weights for minimal feedforward nets with a given input-output map , 1992, Neural Networks.

[5] Max H. Garzon,et al. Boolean Neural Nets are Observable , 1994, Theor. Comput. Sci..

[6] Rufus Bowen,et al. On axiom A diffeomorphisms , 1975 .

[7] Eduardo D. Sontag,et al. Mathematical control theory: deterministic systems , 1990 .

[8] Robert Hecht-Nielsen,et al. On the Geometry of Feedforward Neural Network Error Surfaces , 1993, Neural Computation.

[9] Eduardo D. Sontag,et al. For neural networks, function determines form , 1992, [1992] Proceedings of the 31st IEEE Conference on Decision and Control.