Neuronales Netzwerk und Fuzzy-Zeitreihen

This work deals with neural networks (NN), particularly multilayer NNs from the perspective of algorithm learning. We will describe the Feed Forward Neural Network (FFNN) and the Recurrent Neural Network (RCNN) and introduce fundamental facts about NNs that will be used later in the dissertation. A neural network is a mathematical model inspired by biological neural networks, aiming to simulate them. It consists of interconnected units - neurons, which are the computational units of a neural network. NNs are part of artificial intelligence. Knowledge is stored in the connections between neurons, known as synaptic weights, which are a simplification of biological dendrites and axons. NN is a universal approximator for the relationships stored in the data - a nonlinear statistical data modeling approximator capable of learning and adapting its structure based on internal and external information conveyed during the learning phase by the NN. For most NNs, it is relatively easy to use them across a wide range of technical and non-technical fields without further theoretical knowledge. There are several NNs that require knowledge to implement and to use the correct set of initialization parameters.