应用加权模糊自适应共振人工神经网络进行模式识别

PATTERN RECOGNITION BASED ON WEIGHTED FARTl

针对ART1神经网络只能解决二值逻辑的问题,本文通过数据变换、加权,将其推广至可解决连续分布问题。并对FART的程序流程提出改进,使其理论结构更合理严密。最后,地下核爆炸、天然地震的模式识别结果证实了上述方法的有效性。

Contraposing to the problem that ARTl neural network can only be used in solveing two-value logic, we use ARTl to solve continuous distribution problems with data transformation and adding weights to them. Furthermore, we ameliorate the program flow of ARTl and therefore the construction of the theory is more reasonable and more rigorous. Finally, the results of classifying the underground nuclear explosion and natural earthquake are presented.