Essential Mathematics for Machine Learning

What the lectures build

This 60-lecture NPTEL course from IIT Roorkee works through the linear algebra that machine learning relies on. It starts with vectors and matrix algebra, then moves to vector spaces, subspaces, bases and linear transformations. The middle section covers norms, orthogonal complements, projections, eigenvalues and special matrices. The final stretch focuses on spectral decomposition, singular value decomposition and low-rank approximation, with Python implementations of the last two.

Who it suits

Learners who want the structure of linear algebra and how it feeds into machine learning, rather than a purely computational treatment, will benefit most. The route from vector spaces to low-rank approximation is a clear one for anyone who wants to understand dimensionality reduction.

Before you begin

• The lecture list is theory-heavy, so pair it with practice problems if you learn by doing.
• The Python sessions are the bridge between the abstract material and code.