Modular neural networks for function approximation

Taek Kwon, Hui Cheng, Michael Zervakis

Research output: Contribution to conferencePaperpeer-review

2 Scopus citations


In this paper we consider a network that consists of two modules of neural networks, and attempt to establish an optimal utilization of these modules. Such a study is useful for developing efficient modular neural network (MNN) architectures. We propose three modular architectures that implement a complementary relation, in which the deficiency of one module is complemented by the other module. We provide simulation examples for these three architectures.

Original languageEnglish (US)
Number of pages6
StatePublished - 1994
EventProceedings of the Artificial Neural Networks in Engineering Conference (ANNIE'94) - St. Louis, MO, USA
Duration: Nov 13 1994Nov 16 1994


OtherProceedings of the Artificial Neural Networks in Engineering Conference (ANNIE'94)
CitySt. Louis, MO, USA


Dive into the research topics of 'Modular neural networks for function approximation'. Together they form a unique fingerprint.

Cite this