Title :
Noninvertibility in neural networks
Author :
Rico-Martinez, Ramiro ; Kevrekidis, Ioannis G. ; Adomaitis, Raymond A.
Author_Institution :
Dept. of Chem. Eng., Princeton Univ., NJ, USA
Abstract :
A method for assessing certain validity aspects of predictions made by neural networks used to approximate continuous (in time) dynamical systems is presented. This method searches for noninvertibility (nonuniqueness of the reverse time dynamics) of the fitted model, an indication of breakdown of proper dynamical behavior. It is useful for computing bounds on the valid range of network predictions
Keywords :
filtering and prediction theory; neural nets; dynamical behavior; dynamical systems; fitted model; network predictions; neural networks; nonuniqueness; reverse time dynamics; validity aspects; Chemical engineering; Computational modeling; Computer networks; Differential equations; Educational institutions; Electric breakdown; Intelligent networks; Neural networks; Nonlinear equations; Predictive models;
Conference_Titel :
Neural Networks, 1993., IEEE International Conference on
Conference_Location :
San Francisco, CA
Print_ISBN :
0-7803-0999-5
DOI :
10.1109/ICNN.1993.298587
