AN INTRODUCTION TO ARTIFICIAL NEURAL NETWORKS Carsten Peterson and Thorsteinn R\"{o}gnvaldsson Abstract: A general introduction to artificial neural networks is given assuming no previous knowledge in the field. Properties of the multilayer perceptron, feature maps, the Hopfield model and the Boltzmann machine are discussed in some detail. Also novel methods of finding good solutions of difficult optimization problems with feed-back networks and so-called elastic nets are described. Throughout the lectures practical hints on how to use the algorithms are given. Potential hardware implementations, both VLSI and optical, are briefly mentioned. The power of the artificial neural network approach is illustrated in three high energy physics applications - quark jet tagging, mass reconstruction and track finding. LU TP 91-23; CERN Yellow Report 92-02, 113-170. Lectures given at the 1991 Cern School of Computing, Ystad, Sweden, August 1991