# Optimization

> ML Prep practice set.

- Stable ID: `optimization`
- Area: machine-learning
- Timebox: 60 minutes

Relate stationary points, convexity, objectives, regularization, and training diagnostics.

## Instructions

1. Separate objective geometry from optimizer behavior.

## Ordered items

1. [Learning as optimization](https://mlprep.iwase.dev/item/mit-6-036-s2019-midterm-q5/?set=optimization) — `mit-6-036-s2019-midterm-q5` (medium, 18 min)
2. [Lasso versus ridge penalties](https://mlprep.iwase.dev/item/original-ml-regularization/?set=optimization) — `original-ml-regularization` (easy, 6 min)
3. [Gradient descent on a quadratic](https://mlprep.iwase.dev/item/original-ml-gd/?set=optimization) — `original-ml-gd` (easy, 7 min)
4. [Reducing elastic net to lasso](https://mlprep.iwase.dev/item/pml-book1-ex11.4/?set=optimization) — `pml-book1-ex11.4` (hard, 15 min)

## Completion

Diagnose whether a problem comes from the objective, step size, or model complexity.

## Next recommended practice

- [Neural Networks](https://mlprep.iwase.dev/set/neural-networks.md)