机器学习—线性回归的梯度下降

Training set:

x = “input” variable
      feature
y = “output” variable
      target
m = number of training examples
(x, y) = single training example
(xⁱ, yⁱ) = i^th training example
ŷ :estimated

cost function

Model : f(x) = wx + b
w, b : parameters
         coefficients
         weights
ŷⁱ = f(xⁱ) = wxⁱ + b
cost function :J(w, b) = $\frac{1}{2m}[{\sum }_{i=1}^m({\hat{y}}^i-y^i{)}^2]$
goal :minimize J(w, b)

gradient descent

outline: start with some w,b
             keep changing w,b to reduce J(w,b)
             until settle at or near a minimum
w = w - α $\frac{ə}{əw}$J(w,b)
b = b - α $\frac{ə}{əb}$J(w,b)
α: learning rate/ the size of step