BP network has been extensively applied.Nevertheless,under the condition of numerous learning samples and complicated input-output mapping relationship,the learning speed is rather low.Based on analysing characteristics of error curved surface of the network,the authors have advanced the rapid BP algorithm,which can greatly raise the learning speed.
[1] Moon W M,et al.Reservoir characterization using feedforward neuralnetworks.63rd Ann,Int.SEG Mtg.Expanded Abstracts,1993.258~263
[2] Lippman R P.An introduction to computing with neural nets.IEEE ASSP Magazine,1987,4:4~22
[3] Amari S I.Mathematical foundations of neurocomputing.Proceedings of the IEEE,1990,78(9)
[4] Mclormack M D.Neural computing in geophysics.In:The leading edge of exploration,1991.11~15
[5] 焦李成,保铮.同伦BP网络理论与算法.电子学报,1993,21(11):1~6
[6] 包约翰.自适应模式识别与神经网络.北京:科学出版社,1992