
上QQ阅读APP看书,第一时间看更新
Understanding the structure of neural networks
The atomic computational unit of a neural network is the artificial neuron. It simulates different basic functions of the biological neuron; it evaluates the intensity of each input, adds the different input values, and compares the result with an appropriate threshold. Finally, it determines what the value of the output is. The characteristic of the neuron to add algebraically the values of its input, represents the first function implemented by the artificial neuron; the sum function. The following diagram shows a scheme of a single neuron with the threshold of activation:

The values acquired as input subjected to an appropriate transformation are then returned as output, as a prediction of the result of the observation.