MML-12-09: Second-Order Methods: Newton-Raphson, Quasi-Newton (BFGS / L-BFGS) & Natural Gradient – Practice Quiz

MML-12-09: Second-Order Methods: Newton-Raphson, Quasi-Newton (BFGS / L-BFGS) & Natural Gradient. 40 independently reviewed questions, including single-select and multiselect MCQs. Cognitive scope: Tier 2–4. Prerequisites: QS 12.8, Part 06.

Learning goals: Derive Newton directions from quadratic models.; Calculate Newton steps and local convergence orders.; Diagnose indefiniteness, singularity and damping.; Distinguish Hessian from inverse-Hessian secant equations.; Apply and verify BFGS curvature updates.; Explain limited-memory storage and two-loop recursion.; Compute natural-gradient directions using Fisher geometry.; Compare scaling, computation and optimizer assumptions.

Progress from conceptual checks to worked calculations, ML diagnostics and derivations within the assigned scope. Every question includes content-based feedback for every choice and authoritative references. Select every correct choice when marked multiselect.

Photo by Yoss Traore on Pexels. Licensed under the Pexels License: https://www.pexels.com/license/.