NEURAL NETWORKS

Neural networks are automated structures, which are caused by the example of natural sensory chains that form (set up) the animal. In other words, Neural networks in development are sets of algorithms simulating human brain activity. That sort of structure study to complete the assignments, by looking through the samples, without being planned out.

Neural Networks comprise a series of the united elements or nodes named artificial neurons that easily imitate themselves in the organic substance. Each contact can transfer one signal from one AN to different ones. The imitation neuron that gets the indicators may convert it and gives signs to neurons united with it.

The signal in the contact among neurons may be a real sum. The finished product is generated by the set of the non-continuous functions of some original material. The layers are the basement of the artificial neurons.

The primary aim of the NN is to settle the problems in the same approaches the personal brain does.

Sensory systems can be applied for:

There are many samples of the NNs, every of which requires functional utilization and the levels of complications.

The most used kind of NN is a feedforward NN (one way from the beginning till the end).

The second kind is a recurrent NN (many ways of transitions).

The third NN is convolutional.