若用线性函数作为神经元激活函数则无法处理复杂的非线性问题。
激活函数在神经网络中的作用
相当于每个神经元都在进行对率回归
学习率控制着梯度下降的搜索步长,学习率过大收敛过程容易发生振荡,学习率过小收敛速度过慢
https://blog.csdn.net/victoriaw/article/details/78075266
https://blog.csdn.net/snoopy_yuan/article/details/70846554
https://blog.csdn.net/victoriaw/article/details/78078027
https://blog.csdn.net/victoriaw/article/details/78075789
https://blog.csdn.net/snoopy_yuan/article/details/71093910
https://blog.csdn.net/victoriaw/article/details/78086636
https://blog.csdn.net/snoopy_yuan/article/details/71703019