4. Theoretical consideration

The concept of neural networks was introduced by [31]. The neural-network approach, also referred to as 'connectionism' or 'paralleled distributed processing', Nonlinear Evapotranspiration Modeling Using Artificial Neural Networks DOI: http://dx.doi.org/10.5772/intechopen.81369

adopts a "Brain metaphor" of information processing. Information processing in a neural network occurs through interactions involving large number of simulated neurons. A neural network (NN) is a simplified model of the human brain nervous system consisting of interconnected neurons in a parallel distributed system, which can learn and memorize the information. In NN, the interneuron connection strengths, known as 'synaptic weights' are used to store the acquired knowledge [32]. In other words, ANN discovers the relationship between a set of inputs and desired outputs without giving any information about the actual processes involved; it is in essence based on pattern recognition. ANNs consist of a number of interconnected processing elements or neurons. How the inter-neuron connections are arranged determines the topology of a network. A neuron is the fundamental unit of human brain's nervous system that receives and combines signals from other neurons through input paths called 'dendrites'. Each signal coming to a neuron along a dendrite passes through a junction called 'synapse', which is filled with neurotransmitter fluid that produce electrical signals to reach to the soma or cell body where processing of the signals occurs [16]. If the combined input signal after processing is stronger than the threshold value, the neuron activates, producing an output signal, which is transferred through the axon to the other neurons. Similarly, ANN consists of a large number of simple processing units called neurons (or nodes) linked by weighted connections. A comprehensive description of neural networks was presented in a series of papers [33–35], which provide valuable information for the researchers.
