TITLE: Regularization of Neural Networks AUTHIORS: Jan Larsen, Lars Kai Hansen Department of Mathematical Modelling, Technical University of Denmark, Building 305, DK-2800 Lyngby, Denmark. Emails: jl,lkhansen@imm.dtu.dk. Claus Svarer Department of Neurology, National University Hospital, DK-2100 Copenhagen O, Denmark. Email: csvarer@pet.rh.dk. ABSTRACT: Neural networks are flexible tools for nonlinear function approximation and by expanding the network any relevant target function can be approximated. The risk of overfitting on noisy data isof major concern in neural network design. By using regularization, overfitting is reduced, thereby improving generalization ability on future data. In this contribution we present a scheme for estimation of regularization parameters using a simple validation set approach. Further work on empirical methods for optimization of neural networks models can be found in [jl-Larsen4]. Appears in Procedings of the Fourth Interdisciplinary Inversion Worksop, Technical University of Denamrk, Denmark, Sept. 24, 1996.