Skip to content
ml-recall

#rotation

3 questions from Optimization & Calculus, Convolutional Neural Networks

Timed test

3 questions · 3 min · score and explanations at the end

Take a test
Q 1 / 3Score 0 / 0

Question 1: Short answer

hardAI-checkedOptimization & Calculus

For f(x)=0.1x12+2x22f(\mathbf{x}) = 0.1x_1^2 + 2x_2^2 the Hessian is diag(0.2,4)\textrm{diag}(0.2, 4), and diagonal preconditioning brings the condition number from 20 down to 1. Rotating the problem by 45∘45^\circ (and doubling it, which does not change any condition number) gives f(x)=0.1(x1+x2)2+2(x1−x2)2f(\mathbf{x}) = 0.1(x_1 + x_2)^2 + 2(x_1 - x_2)^2. What is the condition number of D−1/2HD−1/2D^{-1/2}\mathbf{H}D^{-1/2}, with D=diag(H)D = \textrm{diag}(\mathbf{H}), for the rotated function?

a number