Learning the rulesIt is a concept in neural network models, which means that the weights in the network will adjust over time and is generally regarded as a dynamic rule on a long time scale.
In general, the learning rule depends on the excitation value of the neuron, and it may also depend on the target value and current weight value provided by the supervisor.
For example, in a neural network used for handwriting recognition, there is a set of input neurons that are stimulated by the data of the input image. After the stimulation values are weighted and passed through a function, the stimulation values of these neurons are passed to other neurons. This process is repeated until the output neuron is stimulated. Ultimately, the stimulation value of the output neuron determines the recognized letter.