Spectral technique for hidden layer neural network training

Autor:	Terry Windeatt, Robert Tebbs
Rok vydání:	1997
Předmět:	Theoretical computer science Artificial neural network Training (meteorology) Phase (waves) Hyperplane Artificial Intelligence Constructive algorithms Signal Processing Computer Vision and Pattern Recognition Limit (mathematics) Hidden layer Algorithm Software Mathematics
Zdroj:	Pattern Recognition Letters. 18:723-731
ISSN:	0167-8655
DOI:	10.1016/s0167-8655(97)00079-2
Popis:	We propose a new constructive algorithm for learning binary-to-binary mappings. Weight constraints derived from a spectral summation are used to check separability during the partitioning phase, and to limit hyperplane movement during training.
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_________::247a1d143f188e1eed1d9cf46f1fe7b6 https://doi.org/10.1016/s0167-8655(97)00079-2 Zobrazit plný text záznamu Full Text from ScienceDirect