Start with arithmetic and fractions. Finish with the calculus, linear algebra, probability and statistics that modern ML is built on. Every idea is built from the one before it.
Two of the most important lessons in the course are free — no sign-up. Feel how the interactive figures teach before you commit to anything.
Six courses, in order. The amber strand — where each idea reappears in machine learning — runs through every lesson.
Start from zero — numbers, fractions, percents, exponents, algebra, graphs, functions.
Sequences, limits, derivatives, optimization, integration, Taylor series.
Vectors, matrices, eigenvalues, SVD, projections — the native language of neural nets.
Multivariable: gradients, Jacobians, Hessians, multivariate optimization — backprop math.
Random variables, distributions, information theory, limit theorems.
Estimation, MLE, hypothesis testing, regression, model evaluation.
One focused page. A plain explanation and worked examples — no walls of notation.
Drag the interactive figures — like the one above — until the idea clicks.
Auto-graded problems with full solutions, generated fresh every time you return.
If you can count, you can begin. Open the first lesson and build all the way up.