Each neuron in the network receives the input from the previous level and outputs to the next level, and there is no feedback in the network, which can be represented by a loopless graph. This network realizes the transformation of signal from input space to output space, and its information processing ability comes from the compounding of simple nonlinear functions. The network structure is simple and easy to implement. The BP network is a typical feed-forward network.<br>(2) Feedback network: there is feedback between neurons within the network, can be represented by an undirectional complete diagram. This kind of neural network information processing is the transformation of state, which can be processed by dynamic system theory. The stability of the system is closely related to the associative memory function. Hopfield networks and Boltzmann machines fall into this category.
正在翻译中..