A systematic synthesis procedure for feedforward neural networks by using the GRBF (generalized radial basis function) network technique
A. Miyazaki, Takayuki Yamada
Abstract
A. Miyazaki, Takayuki Yamada
Abstract
Some of the problems that one should first be able to address when developing a synthesis procedure for feedforward neural nets are considered to be the following: development of a systematic approach to the representation of neural networks, that is, choosing the number of hidden layers and the number of units in each hidden layer required to achieve a given level of performance in a given applications; and development of a systematic procedure for the learning of neural networks, that is, setting the weights of a feedforward neural network by using much of the information contained in a given set of examples of input-output pairs. This paper deals with the two problems above by using the GRBF (generalized radial basis function) network technique closely related to approximation techniques such as generalized splines and regularization theory, and aims to offer a framework within which it is possible to address the problems and provide a systematic synthesis procedure for feedforward neural networks.
OpenAlex reports 1 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Some of the problems that one should first be able to address when developing a synthesis procedure for feedforward neural nets are considered to be the following: development of a systematic approach to the representation of neural networks, that is, choosing the number of hidden layers and the number of units in each hidden layer required to achieve a given level of performance in a given applications; and development of a systematic procedure for the learning of neural networks, that is, setting the weights of a feedforward neural network by using much of the information contained in a given set of examples of input-output pairs. This paper deals with the two problems above by using the GRBF (generalized radial basis function) network technique closely related to approximation techniques such as generalized splines and regularization theory, and aims to offer a framework within which it is possible to address the problems and provide a systematic synthesis procedure for feedforward neural networks.
Key concepts: Feedforward neural network, Feed forward, Artificial neural network, Computer science, Representation (politics), Radial basis function, Regularization (linguistics), Time delay neural network