Neural networks and applied hydrogeology

Neural networks are mathematical black – box models which try to reply functions and form of human brain neurons. Development of neural networks started after the World War II, when first studies on artificial neurons were published (McCulloch, Pitts, 1943). The learning ability, as in the human neurons, was introduced in 1949 (Von Neumann, 1949). The most known form of neural network, the perceptron, was developed at the end of ‘50s (Rosenblatt, 1958), and its aspect is stayed until nowadays which few changes. Obviously, the development of modern computers allowed to improve and fine the learning techniques and the structure of model, creating the most utilized algorithm, the error backpropagation (Rumelhart, Hilton, Williams, 1986).