Skip to content
ml-recall

#sgd

18 questions from Optimization & Calculus, Linear Models & Regularization, Logistic Regression & Softmax, Training Deep Networks, PyTorch

Timed test

18 questions · 18 min · score and explanations at the end

Take a test
Q 1 / 18Score 0 / 0

Question 1: Single choice

hardAI-checkedOptimization & Calculus

For a convex objective, stochastic gradients bounded by LL and a start at distance rr from the optimum, SGD with step sizes ηt\eta_t satisfies E[R(xˉ)]−R∗≤r2+L2∑t=1Tηt22∑t=1Tηt.E[R(\bar{\mathbf{x}})] - R^* \le \frac{r^2 + L^2 \sum_{t=1}^T \eta_t^2}{2\sum_{t=1}^T \eta_t}. Which schedule makes this bound go to 0 as T→∞T \to \infty?

Choose one option