Summary of Neural Network Operation

Second, the PEs are connected to each other in a particular fashion. In our example each of the input PEs is connected to both HL PEs. In contrast to this, each of the HL PEs is connected to only one output layer PE. There are different types of ANN, and each has its own connection scheme that specifies which PEs are connected to each other. 

each layer has a certain number of PEs. In our example each of the three layers happens to have two PEs. The number of PEs in each layer depends on the type of network and the data being used. 

Real-life data that is interesting is not as trivial as the example above. But much data does exist that can be split into a set of independent variables (the inputs to an ANN) and a set of dependent variables (the desired outputs of an ANN). Suppose you have a data set with five independent variables and one dependent variable, and you know through some combination of experiment and computation the values of all six variables for, say, 50 cases. [In our exam- 

you want to use your trained network to predict the dependent variable value for the 51st case—a pattern that is not in the training set and consists of a value for each of the five independent variables. Simply apply these five inputs to the input PEs and observe the output at the output PE. This is the predicted value. In this phase of operation the network is in a test mode or recall mode, as opposed to a training mode. 

At this point we must address two questions What happens if the network cannot learn the training data And, how good (accurate, reliable) are the predicted results In response to the first question, we note that usually a network cannot or will not completely learn the training data Very often, this turns out to be inconsequential because as a rule there is experimental error or uncertainty associated with data, both input and output. Also, it often happens 

---