一： what is Activation Function？
它仅仅是一个函数
二：why we use Activation function with Neural Networks?
it map the resulting values in betwwen 0 to 1 or -1 to 1
**函数大体可以分为两类：线性和非线性
线性函数的输出一般不受限制，他不能处理复杂多变的神经网络参数。所以**函数一般都是非线性的，它让模型可以很容易泛化或者适应各种数据，并区分输出。
三 Sigmoid function
sigmoid looks like S-shape
the main reason why we use sigmoid function is because it exists between 0 to 1,因此它很适合 predict the probability, since probability of anything exists only between 0 to 1
this function can cause a neural network to get stuck at the training time,the softmax function is a more function which is used for multiclass classfication.
四： tanh
the advantage is that the negative inputs will ba mapped strongly negative and the zero inputs will be mapped near zero .
五：ReLU(Rectified Linear Unit)
Relu is the most used activation function in world right now ,
Range:[0 ,infinity)