Dr. Pei

Electrical and Computer Engineering | STCO

Menu

Compute Backpropagation Derivatives

Consider two features x₁, x₂ for a single training set:

${\color{Magenta} z=w_{1}^{T}x_{1}+w_{2}^{T}x_{2}+b}$

${\color{Magenta} \hat{y}=a=\sigma (z)=\frac{1}{1+e^{-z}}}$

${\color{Magenta} L(a,y)=-[ylog(a)+(1-y)log(1-a)]}$

Last posts

Sidebar

×

Google Scholar