Artificial neural networks (KNN), Artificial Neural Networks (ANN) or Simulated Neural Networks (SNN), are computer models that are based on the way humans think and emulate the information processing, information storage and learning process of the human brain.
Neural networks replicate certain capabilities of the brain. For example, pattern recognition or the ability to view decisions and errors inrelation to data. A neural network is a kind of software- hardware model of the brain in which simple decisions are linked to form a system. In these decisions, where there is no linear relationship between input and output variables, complex decision processes take place. Whereby the characteristics of the neurons occurring in the human brain are reproduced by the corresponding arrangement of the individual network nodes.
Neural networks consist of the connections of fully and partially meshed neurons. The output of one neuron represents the input of the following one up to the result output. Each neuron within the neural network has one or more inputs and produces a working output from them. Each input is given a weighting factor by which the input values of the neurons are modified. The neuron manipulates the input and the result on a mathematical basis. The network learns from positive and negative experiences, where the weighting factors, through human intervention or through predetermined algorithms, bring the results closer to known results. This knowledge is stored in the intermediate connections by the neurons. Such a neural system bears some resemblance to children who learn from experience and practice.
The key advantages of neural networks are that these networks can interpret and represent linear and nonlinear relationships and that they learn these relationships directly from the data being modeled. Therefore, neural networks can be used for predictive systems and control systems. It can process certain tasks much more effectively than other computer systems.
A Convolutional Neural Network( CNN) represents a special form of an artificial neural network.