基于模糊聚类方法的神经元形态分类识别

神经元形态分类识别是"人类脑计划"研究首要解决的问题。神经元真实形态复杂多样,利用物理观察和日常经验无法进行分类识别,传统的分类识别算法难以解决形态相似的神经元分类识别的误判现象。针对神经元形态分类误判与类别重叠问题,提出神经元几何形态特征提取方法,设计神经元形态特征自由分类模型,从而为神经元的精确分类、有效识别与新型命名提供方法支持和实践参考。实验结果表明,该分类模型具有较高的运行效率和聚类精度,较好地解决了分类误判和类别重叠问题。

Classification and Identification of the morphology of Neurons Based on Fuzzy Clustering Method

The classification and identification of the morphology of neurons is the primary problems of "Human Brain Project". For the real morphology of neurons are complex and diverse, the use of physical observation and everyday experience can not be used for classification, and the traditional algorithms of classification and identification are difficult to solve problems of the phenomenon of false positives to classification of the similar neurons. Aiming at the false positives to classification and categories overlap problems of the morphology of neurons, The paper proposes the feature extraction method of neurons geometric shape, and designs a new free classification model of neurons morphology characteristics, which provides method support and practical reference for the precise classification and new naming of neurons .The experimental results show that the classification model has higher operation efficiency and clustering precision, provides a good solution to the problem of false positives and categories overlap.