Functional Network and Tunable Activation Function Neural Network

Functional network is a recently introduced extension of neural networks. Unlike neural networks, it deals with general functional models instead of sigmoid-like ones. And in these networks there are no weights associated with the links connecting neurons. In this paper, firstly, the architecture of functional network is deformed, compares the structure of networks and learning model algorithm, and approximates performance with tunable function neural network. It is pointed out that people study of with tunable function neural network is only special case, and functional network is more generally model. Numerical analyses results show that better convergence performance of functional network algorithm over tunable function neural network