Many practical applications of neural networks require the identification o
f strongly non-linear (e.g., chaotic) systems. In this paper, locally recur
rent neural networks (LRNNs) are used to learn the attractors of Chua's cir
cuit, a paradigm for studying chaos. LRNNs are characterized by a feed-forw
ard structure whose synapses between adjacent layers have taps and Feedback
connections. In general, the learning procedures of LRNNs are computationa
lly simpler than those of globally recurrent networks. Results show that LR
NNs can be trained to identify the underlying link among Chua's circuit sta
te variables, and exhibit chaotic attractors under autonomous working condi
tions. (C) 2001 Elsevier Science Ltd. All rights reserved.