Neural Networks(2)-Gradient desc
2017-10-30 本文已影响0人
MWhite
17/10/30 MWhite's learning notes
Recap
TIM截图20171030220137.jpg
Gradient descent algorithm
want to find w&b that minimize J(w,b)
w:=w-a*dJ(w,b)/dw
(a-learning rate)
Derivate&calculus
Skip
Computation Graph
TIM截图20171030225155.jpg
TIM截图20171030230314.jpg
