论文部分内容阅读
针对一般模糊极小极大神经网络在处理重叠超盒和包含超盒时,出现新的类而标识为未知类,进而无法达到聚类预期效果的问题,提出了通过超盒的收缩过程来加入新类或删除一个已存在类的一般模糊极小极大神经网络,它继承了一般模糊极小极大神经网络的优点,并且避免了一般模糊极小极大神经网络在分类时的随意性,弥补了一般模糊极小极大神经网络无法达到聚类预期效果的目的,以及提高了模式分类的准确性和高效性.最后,通过实例验证了方法实用有效.
Aiming at the problem of general fuzzy minimal maximal neural networks dealing with overlapping hyperboxes and containing hyperboxes, a new class is identified as an unknown class, and then the desired effect of clustering can not be achieved. It is proposed to add New class or delete an existing class of general fuzzy minimal maxima neural network, which inherits the advantages of general fuzzy max maxima neural network, and avoids the randomness of general fuzzy minimax neural network in classification, Which can make up for the purpose that general fuzzy maximal maximal neural network can not reach the expected result of clustering and improve the accuracy and efficiency of pattern classification.Finally, the example is used to verify the method is practical and effective.