人工神经网络用于酸性染料的分类

The Classification of Acidic Dyes with Artificial Neural Network

提出用ExtendedDelta－Bar－Delta（简称EDBD）网络对酸性偶氮染料进行分类，网络结构为4－6－5，并对网络结构进行了优化．一次分类结果与采用GCEDM[1]逐次分类的结果很好地吻合，采用EDBD网络分类，比采用GCEDM分类法简单、快速、准确．

The acidic dyes were classified by using Extented Delta-Bar-Delta (EDBD). Thebest structure of network was 4-6-5. The optimized learning times is about 5000. It isdifficult to classify these dyes because their structures are very similar. Compared with GCEDMand other methods which were applied formerly, the EDBD method have the advantanges ofmore stable classification standard, fewer parameters and quicker velocity.