全部 标题 作者
关键词 摘要

OALib Journal期刊
ISSN: 2333-9721
费用:99美元

查看量下载量

相关文章

更多...

具有更严格警戒测试准则的ART2神经网络

DOI: 10.11834/jig.20010121

Keywords: 模式识别,神经网络,ART2,警戒测试准则

Full-Text   Cite this paper   Add to My Lib

Abstract:

在ART2神经网络的标准警戒测试准则中,通过引入截断双曲线函数来计算输入矢量与神经网络由顶向下权重矢量之间的相似程度,而提出了一种新的具有更严格警戒测试准则的ART2神经网络。截断双曲线函数一方面抑制输入样本中的噪声,另一方面,如果输入矢量某些分量与由顶向下权重矢量对应分量之间存在冲击变化时,则截断双曲线函数将放大这些对应分量之间的冲击变化。而且这种新的警戒测试准则具有更强的抗噪声能力。即在较低的输入信噪比水平上,具有更严格警戒测试准则的ART2神经网络比标准ART2神经网络具有更高的正确识别率。

References

[1]  Baxter R. Error propagation and supervised learning in adaptive resonance networks. In,International Joint Conference on Neural Networks Ⅱ, Piscataway, NJ. USA, 1991,423-429.
[2]  Huang J. Georgiopoulos M, Heileman G L. Fuzzy AET properties. Neural Networks. 1995,3:202-213.
[3]  Carpenter G A,Grossberg S. ART2:Stable self- organization of category recognition codes for analog input patterns. Applied Optics, 1987,26.4919- 4930.
[4]  Li F, Zhan J. Fuzzy adapting vigilance parameter of ART-Ⅱ neural nets. In: IEEE International Conference on Neural Networks, Orlando, FL, USA, 1994,3:1680-1685.
[5]  Ming L, Paul S W. Personal identification by palm prints recognition. In:The 10th International FLAIRS Conference.Daytona Beach, Florida, USA, 1997,1211-1218

Full-Text

Contact Us

service@oalib.com

QQ:3279437679

WhatsApp +8615387084133