Welcome to our site! EDAboard.com is an international Electronic Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.
Neural network is a kind of Artificial Intelligence. The basic elements in NN are neurons, weights, and threshold.
NN is a model simulating human brain, where it can learn by training. The concept of NN is not complicated, you can learn from many books. There are several ebooks about NN in EDAboard Ebook Upload/Downlaod.
There're many architectures and training methods for NN, where each of them has strength and weakness respectively. Looking for a suitable architecture and training algorithm for particular application is very important to make the NN work.
NN is one way of simulating the way humans learn things.
It is a network of nodes (Neurons, Nerves), relating with each other with links (synapses). Each link has a Wait. So we can simulate it with a matrix, each row representing the waits between a certain node and the other nodes. This network can have layers, it means there are two or more sets of nodes, each have a matrix of its own, and there are some links between layers. The first set (which may be the only set) has some links to the outside, representing INPUTS. The final set (which can be the only set, too) has outer links as the OUTPUT of the network. You apply the inputs to the INPUT, and you get the output from OUTPUT. Learning, means changing, in some manner, the waits of the nodes, so that the networks give the right outputs with certain inputs. So, they can give the outputs of unknown inputs, by learning the relation between the INPUTS and OUTPUTS of your certain problem.
There are some different kind of Nets, such as RBF (Radial Basis Functions), Multi-Layer perceptrons (MLP), Kohonen Networks, so on... .
You can find a comprehensive description of NNs here: http://www.cs.stir.ac.uk/~lss/NNIntro/InvSlides.html