ML ATLAS
From Mathematical Foundations to Real-World Intelligence — an open, static, interactive machine-learning textbook for a university ML course.
01 Learning philosophy
Problem first
Each model opens with a real problem and the data it operates on — not a definition. The mathematics arrives only when a question needs it.
Maths with a why
Every formula is tied to the object it explains. No symbol floats without a reason for existing in that model.
See the pattern
Every model has a focused visual explanation: read the chart, follow the shape and connect the mathematics to the prediction.
Choose honestly
No single algorithm is best. The Compass and Decision Engine rank, explain trade-offs, and always say "it depends".
02 What is inside
15 model pages
Regression, classification, clustering and dimensionality reduction — every one with a clear visual learning arc and interactive explanation.
Mathematical Foundations
Linear algebra, calculus, probability, statistics and optimization — each concept with definition → intuition → formula → ML connection.
Tools for choosing
A searchable concept index, a Model Compass with filters, and a step-by-step Model Decision Engine.
Your progress
A lightweight local progress tracker — mark models as learned and watch the progress bar advance. No account needed.
03 Design principles
- Static by design — plain HTML, CSS and vanilla JS. No server, no database, no account. Everything runs in the browser; progress lives in localStorage.
- Visual but truthful — every chart, diagram and explanation is designed to make the algorithm's core idea easy to see.
- Accessible — semantic landmarks, keyboard navigation, focus styles, ARIA labels, reduced-motion support, readable contrast.
- Lightweight — no framework, no build step, no heavy dependencies. MathJax is the only external asset.
04 Built with
To use ML ATLAS, open index.html in a modern browser. No build or install needed; serving the folder over a static file server keeps ES module imports happy.
