论文部分内容阅读
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。
ART1 neural network can only solve the binary logic problem, this paper through data transformation, weighting, it can be extended to solve the problem of continuous distribution. The improvement of FART’s program flow makes the theoretical structure more reasonable and rigorous. Finally, the results of pattern recognition of underground nuclear explosions and natural earthquakes confirm the validity of the above method.