Training trajectories by continuous recurrent multilayer networks

L Leistritz; M Galicki; H Witte; E Kochs

doi:10.1109/72.991415

Training trajectories by continuous recurrent multilayer networks

IEEE Trans Neural Netw. 2002;13(2):283-91. doi: 10.1109/72.991415.

Authors

L Leistritz¹, M Galicki, H Witte, E Kochs

Affiliation

¹ Inst. of Med. Statistics, Comput. Sci. and Documentation, Friedrich-Schiller-Univ., Jena.

PMID: 18244431
DOI: 10.1109/72.991415

Abstract

This paper addresses the problem of training trajectories by means of continuous recurrent neural networks whose feedforward parts are multilayer perceptrons. Such networks can approximate a general nonlinear dynamic system with arbitrary accuracy. The learning process is transformed into an optimal control framework where the weights are the controls to be determined. A training algorithm based upon a variational formulation of Pontryagin's maximum principle is proposed for such networks. Computer examples demonstrating the efficiency of the given approach are also presented.