首页> 外文会议>International symposium on neural networks;ISNN 2009 >Geometry Algebra Neuron Based on Biomimetic Pattern Recognition
【24h】

Geometry Algebra Neuron Based on Biomimetic Pattern Recognition

机译:基于仿生模式识别的几何代数神经元

获取原文

摘要

Biometric Pattern Recognition aim at finding the best coverage of per kind of sample's distribution in the feature space. It is based on the analysis of relationship of sample points in the feature space. According to the principle of "same source", research the same kind of samples' distribution in the feature space can get eigenvector information with low data amount. This can be realized by 'coverage recognizing method of complex geometric body in high dimensional space'. Self-adaptive topological structure of high dimensional geometrical neuron model offers theoretical basis for its realization. But it has been investigated in 2D sample space. In this paper ,we extend to Multispectral Image sample space by Clifford Algebra,and propose geometry algebra neuron by biomimetic pattern recognition theory. The experiment result proves the efficiency of our theory.
机译:生物特征模式识别旨在寻找特征空间中每种样本分布的最佳覆盖范围。它基于对特征空间中样本点之间关系的分析。根据“同一源”原理,研究特征空间中相同种类的样本分布可以获得较少数据量的特征向量信息。这可以通过“高维空间中的复杂几何体的覆盖识别方法”来实现。高维几何神经元模型的自适应拓扑结构为其实现提供了理论依据。但是已经在2D样本空间中对其进行了研究。本文通过Clifford代数扩展到多光谱图像样本空间,并通过仿生模式识别理论提出了几何代数神经元。实验结果证明了本文理论的有效性。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号