基于人工神经网络的乳腺癌诊断模型

Diagnosis Model Based Neutral-network in Galactophore Cancer Cell Identification

针对具体的乳腺癌诊断分类问题 ,研究了多层径向基函数 (RBP)网络的分类机理和初始化优化参数 ,采用动量法和学习率自适应调整两种策略方法 ,建立了基于BP神经网络和径向基 (RBF)神经网络的乳腺癌两种诊断模型。讨论了径向基神经网络的分类机理 ,同时对数据作了预处理。径向基(RBF)神经网络具有较强的非线性并行处理能力和容错能力。仿真结果表明 ,所设计的RBF网络模型性能稳定 ,训练时间短 。

To the classified problem of the Galactophore Cancer Cell Identification, the paper researched the classified mechanism and optimized parameter for more-layer radial basis function (RBF) network, adopted the method of gradient descent with momentum and adaptive learned backpropagation, constructed the two-model of the Galactophore Cancer Cell Identification based on BP Neural-Network and RBF Network. Classified principle based on RBF Neural network was also discussed, and the method for data processing was studied. RBF network has more advantage, such as fault tolerance, nonlinear mapping, etc.. Experiment results show that, on the model based RBF Neural Network, performance is steady, training time is short and classified results are good.