Title :
A new supervised learning method of neural networks and its application to the land cover classification
Author :
Ohkubo, A. ; Niijima, Koichi
Author_Institution :
Fukuoka Inst. of Health & Environmental Sci., Dazaifu, Japan
Abstract :
Proposes a new learning method of three-layer neural networks based on the concept of domains of recognition in the input space. The neural network learns by minimizing a cost function which is derived from geometrical properties of the domain. This learning method is applied to land cover classification problems for multi-band data
Keywords :
learning (artificial intelligence); neural nets; terrain mapping; cost function; geometrical properties; land cover classification; multi-band data; supervised learning method; three-layer neural networks; Biomedical informatics; Cost function; Gaussian distribution; Learning systems; Neural networks; Pattern recognition; Remote sensing; Supervised learning; Training data; Urban areas;
Conference_Titel :
Geoscience and Remote Sensing Symposium, 1999. IGARSS '99 Proceedings. IEEE 1999 International
Conference_Location :
Hamburg
Print_ISBN :
0-7803-5207-6
DOI :
10.1109/IGARSS.1999.774633
