ML Crash Course is Google's introductory machine learning course, centered on core concepts: features and labels, loss functions, gradient descent, overfitting, and the metrics used to evaluate classification and regression, with exercises after each section. It covers classical machine learning, not large-model prompting or API calls, and is best for terms and judgment criteria. Pages come in several languages and translations may lag the English original; use the English version as reference.
What it does
- Core concepts such as features, labels, loss and gradients
- Clear definitions for classification and regression metrics
- Exercises after each section for self-check
- Well-defined terminology as a lookup starting point
- Mostly framework-independent and stable
Who it is for
- Stuck on terms like accuracy and recall in someone else's plan
- Wanting machine learning basics, not only LLM phrasing
- A quick terminology pass before a review or interview
- One metric definition for a whole team
How to get started
- Start with the term-heavy sections instead of reading in order
- Do every exercise and note where you were wrong
- Turn the metric definitions into team review language
