praemineo / gradient-descent

This is an implementation problem of gradient descent in which you can also see the descendants of a point on the plane to reach the minima with the help of the graph and can understand the working of gradient.

Gradient Descent

This is an implementation problem of Gradient Descent in which you can also see the descendants of the point on the plane to reach the minima with the help of the graph through which you can understand the working of gradient descent with clarity.

Initialization

current point

x0 = 3

learning rate

lr = 0.01

derivation of funtion

dy/dx = d/dx [(x+5)^2] = 2 * (x + 5)

Iteration 1 :

x1 = x0 - lr * dy/dx x1 = 3 - (0.01) * (2* (3+5)) = 2.84

Iteration 2 :

x2 = x1 - lr * dy/dx x2 = 2.84 - (0.01) * (2* (2.84 + 5)) = 2.68

.

.

.