Skip to content
ml-recall

#l1-regularization

6 questions from Linear Models & Regularization, Training Deep Networks

Timed test

6 questions · 6 min · score and explanations at the end

Take a test
Q 1 / 6Score 0 / 0

Question 1: Short answer

mediumAI-checkedLinear Models & Regularization

You minimize L(w)+λ∑i∣wi∣L(\mathbf{w}) + \lambda\sum_i |w_i| (ℓ1\ell_1 regularization) by gradient descent with learning rate η=0.1\eta = 0.1 and λ=0.2\lambda = 0.2. A weight has value w=0.5w = 0.5, and the data loss gives it the gradient ∂L/∂w=0\partial L / \partial w = 0. What is ww after one update?

a number