Читать книгу Handbook on Intelligent Healthcare Analytics - Группа авторов - Страница 33

Let us switch now to artificial neural network (ANN) for memory. Again, it will have an overview of how these works are done; let the reader study alternative in-depth perspectives, and propose Raul Rojas’ excellent text (1996). The section uses one aspect of a network and learning method that explains how business application vendors use a network, other network types, and learning processes. It is defined as having thousands of neurons, each of which is associated with more than a thousand other neurons in a rather simplified human brain model. Each neuron receives an electrical signal and transmits it to other brain network neurons. The neuron receives a signal from its associated neurons, and does not transmit the signal to other neurons immediately but waits until the concentration of the signal energy reaches level. In general, the brain learns by changing the amount of these connections and the signal thresholds.

ANN is constructed along identical lines except that node collections execute the location of neurons connected in the network, where a three-layer network is shown for ease. It has several layers defined as the input, hidden layers, and output of the neurons forming the interconnection network. The input neurons are the first information to deal with the problem, and the results and the solutions are in the output neurons. The hidden layer is an input and output layer network link. The diagram shows only one hidden layer, and we adhere for simplicity to one layer in this section, while there may be several such layers in some implementations.

The arrows in the image show the link between the neurons input n, the k hidden neurons, and the neurons in output m. Wisdom is seen as being fed on the left to right and is regarded as a feeding process. We undergo a back-breeding process in later portions. The way the network functions by its neurons has two major characteristics:

Neurons receive feedback from other neurons, however, the neuron also “flies” while the added neuron knowledge is of vital importance (a firing threshold). Information passing from one neuron to another is weighed by a variable that does not have a value affected by data within either neuron. The network is used to efficiently define alternatives to the issue by manipulating weighting variables.