摘要
Freeman在大量神经生理实验的基础上建立了生物嗅觉系统的非线性神经网络模型,即K系列模型.其中,KIII模型是一种混沌神经网络,它的模式识别机制与以往的人工神经网络完全不同,更接近实际生物神经系统的工作模式.研究通过对26个英文字符的学习、识别研究得出,系统相对于传统的神经网络有着很强的学习能力,学习5～6次就能有很好的识别能力,在10次达到最优学习效果,并与真实神经系统学习过程中的倒"U"曲线相对应.
Prof. Freeman established a set of nonlinear neural network model based on olfactory system and physiological experiment. KIII model is a chaotic neural network system, offers a new concept and mechanism, which is different from conventional neural network. It is recognized 26 English capital letters successfully when the system studied those patterns for 5 to 6 times. When the system studied 10 times, the best effect occurs, which is corresponded to the Inverted U-Shaped Curve in real neural system learning procession. This results proved the recognition ability of this new method.
出处
《复旦学报（自然科学版）》
CAS
CSCD
北大核心
2004年第5期710-713,共4页
Journal of Fudan University：Natural Science
基金
"九七三"重大基础研究前期研究专项(2002CCA01800)
教育部留学回国人员启动基金资助
