摘要
提出新的非线性映射算法 ,并分别采用传统非线性映射算法和新的非线性映射算法 ,将 8维橄榄油样本映射于平面。其中新的非线性映射算法获得更好保留样本模式拓扑结构的映射平面 ,映射平面清晰地反映模式的类别关系 ,即同类模式都清晰地聚集在一起 ,实现聚类。
Non-linear mapping (NLM) was applied to project high-dimensional complex chemical information down on the two-dimensional plane, while the topology-preserving map was obtained, on which the important relationship among data can be visualized easily. Furthermore, a modified NLM (M-NLM) with a new algorithm of adaptive mapping error and a novel pseudo-similar random optimization approach was proposed to overcome the deficiencies of conventional NLM (C-NLM). Finally, a typical example of mapping eight-imensional olive oil samples onto two-dimensional plane was employed to verify the effectiveness of M-NLM and C-NLM. The results show that the topology-preserving map obtained by M-NLM can well represent the classification of original patterns and is much better than that obtained by C-NLM.
出处
《分析化学》
SCIE
EI
CAS
CSCD
北大核心
2003年第8期928-931,共4页
Chinese Journal of Analytical Chemistry
关键词
复杂化学模式群
非线性映射
拓扑结构
类随机优化算法
聚类
橄榄油
complex chemical pattern
non-linear map
topology-preserving
pseudo-similar random optimization approach
cluster
