Title :
Sensitivity analysis of neural models
Author :
Lamy, D. ; Borne, P.
Author_Institution :
Lab. de Autom., Ecole Centrale de Lille, Villeneuve d´´Ascq, France
Abstract :
This paper investigates the sensitivity of neural models to weights perturbation in a system identification task. Analytical expression for sensitivity is derived from a notation based on Kronecker product and vector valued function of matrix. Experimental results highlight this sensitivity measure when investigating model structure. A comparison with statistical sensitivity results confirms usefulness of our approach. Search for minimum output sensitivity appears to be a nice indicator for proper model order choice
Keywords :
identification; modelling; neural nets; sensitivity analysis; Kronecker product; analytical expression; minimum output sensitivity; model order choice; neural model sensitivity analysis; statistical sensitivity; system identification; vector-valued function; weights perturbation; Art; Delay; Multi-layer neural network; Neural networks; Nonhomogeneous media; Polynomials; Predictive models; Sensitivity analysis; System identification; Vectors;
Conference_Titel :
Systems, Man, and Cybernetics, 1994. Humans, Information and Technology., 1994 IEEE International Conference on
Conference_Location :
San Antonio, TX
Print_ISBN :
0-7803-2129-4
DOI :
10.1109/ICSMC.1994.400055
