Quick Check 17.1 – Data 100, Summer 2020

Which of the following functions are convex? *

f(x) = x^2

f(x) = x^2 + x^4

f(x) = x^3

f(x) = -x^2

f(x) = x^4

Required

Which of the following reasons are convex loss functions favored over non-convex loss functions? *

A local optimum of a convex function is always a global optimum.

The minimizer for a convex loss function always has an analytical solution.

Gradient descent with a convex loss function does not require a stopping criterion (i.e. stopping after 20 iterations).

Required

Gradient descent with a decreasing step size at each iteration will always find the global minimum of a non-convex function. *

True

False