Artificial neural network is networks of artificial neurons. Artificial Neural Network is a branch of artificial intelligence research and it is among the fundamental object of the neuro-computer science. You can go through our previously posted article on Artificial Intelligence, Machine Learning and Autonomic computing.
Artificial Neural Network is an abstraction ( modeling ) of information processing and less about the reproduction of biological neural networks.
How this Artificial Neural Network works
Artificial neural network is mostly based on the networking of many McCulloch-Pitts neurons or slight variations.. In principle also other artificial neurons are used in KNNen – such as the high-order neuron. The topology of a network or the assignment of connections to nodes, must depend on its mission. After the construction of an?Artificial Neural Network in the training phase, in which the network “learns”. In theory, a learning network through the following methods:
---
- Development of new connections, deleting existing Artificial Neural Network connections
- Changing the weights of neurons within an Artificial Neural Network
- Adjusting the thresholds of the neurons
- Adding or deleting neurons within a?Artificial Neural Network or another?Artificial Neural Network supervised by it.
In addition, the learning behavior changes when one change the activation function of neurons or the learning rate of the network. In practical terms, a network “learns” mainly by modifying the weights of the neurons. An adjustment of the threshold can be effected by other neuron. Artificial Neural Network is capable of running complicated nonlinear functions via a “learning” – algorithm, through the iterative or recursive procedure from available input and desired output values of all parameters to learn. Artificial Neural Network is a realization of a paradigm, because the function is composed of many simple homogeneous parts.
Classes and Types of ANN?Artificial Neural Network (ANN)
Using Artificial Neural Network?is not so straightforward and a relatively good understanding of the underlying theory is essential.
Basically, the classes of networks differ mainly by the different network topologies and connection types. For example, layered, layered, feedforward and feedback networks.
- McCulloch-Pitts networks
- Learning matrix
- Perceptron
- Adaline model
- Self-Organizing Maps (including Kohonen networks) (SOM)
- Growing Neural Gas (GNG)
- Learning vector quantization (LVQ)
- Boltzmann machine
- Cascade Correlation networks
- Counterpropagation networks
- Probabilistic neural networks
- Radial basis function networks (RBF)
- Adaptive resonance theory (ART)
- Neocognitron
- Spiking Neural Networks (SNN)
Special form of Artificial Neural Network:
- Pulse Coded Neural Networks (PCNN)
- Time Delay Neural Networks (TDNNs)
- Recurrent neural networks (RNNs)
- Bidirectional associative memory (BAM)
- Hopfield networks
- Elman networks (including simple recurrent network, SRN)
- Jordan networks
- Oscillatory neural network

Application of?Artificial Neural Network
Artificial Neural Network ?has the special feature which make Artificial Neural Network all applications interesting, where little or no explicit (systematic) knowledge exists about the problem to be solved. Artificial Neural Network ?are used in text recognition, image recognition and face recognition.
Obstructive factors for using? Application of?Artificial Neural Network
The training of Artificial Neural Network usually leads to high-dimensional, nonlinear optimization problems.