In the beginning of eighties, J. Hopfield developed a new model of neural network inspired by study by auto-associative networks. During the research, he developed an energy function that has major impact on correct function of the network – rules for learning and relaxation are derived from said function. Hopfield illustrated the application of this network on several physical models. There are several modifications of this network today – Hopfield network mey be used as associative memory, classifier or optimization problem solver. The behavior of the network may be well illustrated on image patterns, as binary values can be easily assigned to image pixels. Hopfield network is not suitable for continuous inputs, as the conversion of continuous signals onto binary values presents a major problem.