Gradient Descent

The Gradient Descent can be given by:

repeat:{

$θ_j:= θ_j−α ∂/∂θ_j J(θ)$

} (simultaneously update for every j=0,1,2….)

On computing the partial differential in the above equation, we get:

repeat:{

$θ_j:=θ_j−α (1/m) ∑_{i=1}^{m}((h_θ(x^{(i)})−y^{(i)})x_{j}^{(i)})$

} (simultaneously update every j=0,1,2…,n)