论文部分内容阅读
为了提高肺癌的诊断效率,建立了一个三角形隶属度函数模糊神经网络。首先用三角形隶属度函数对非二值参数进行模糊化,然后和二值参数一起作为BP神经网络的输入。从117病例中随机选出的73例作训练集,剩下的作测试集。用测试集来测试BP和三角形隶属度函数模糊神经网络的性能,并将二者的结果进行了比较。三角形隶属度函数模糊神经网络的性能明显优于BP神经网络。
In order to improve the diagnostic efficiency of lung cancer, a triangular membership function fuzzy neural network was established. Firstly, the membership function of triangle is used to fuzzify the non-binary parameters, and then the two-valued parameters are used as the input of BP neural network. Seventy-three randomly selected cases from 117 cases were selected as training sets, and the rest were tested. The test set was used to test the performance of BP neural network with triangular membership function and the results of the two were compared. Triangular membership function of fuzzy neural network performance is better than BP neural network.