在RBF神经网络学习过程中，I出F神经元先计算输入与中心之间的距离，然

后再对这一距离进行某种非线性变换。输出层和隐藏层分别完成不同的任务，这两层学习的策略也不相同。输出层是对线性权进行调整，采用的是线性优化策略，

因而学习速度较快。而隐藏层是对传递函数的参数进行调整，采用的是非线性优

化策略，因而学习速度较慢。

RBF算法选用高斯函数作为隐藏层传递函数时，由隐藏层来实现从

x哼R，(x)的非线性映射，由输出层来实现从R,(X)--->y。的线性映射。-In the RBF neural network learning process, I first calculate the F neurons in the input and the distance between the center and then on the distance of a nonlinear transformation. Output layer and hidden layer are different tasks, this is not the same two learning strategies. Linear output layer is the right to adjust, using the linear optimization strategy, thus learning faster. The hidden layer is the transfer function parameters can be adjusted using a nonlinear optimization strategy, and thus learn more slowly. Algorithms use Gaussian function as the RBF hidden layer transfer function from hidden layer to achieve from the x Well R, (x) non-linear mapping from the output layer to transition from R, (X )--->　y. Linear mapping.