Source-coding loop
Use for entropy, prefix codes, Huffman coding, and compression audits.
Learn information theory by calculating, coding, and reasoning
A self-contained course from bits and entropy to compression, noisy channels, inference, and modern machine-learning objectives. The lessons combine short explanations, checkable math, browser labs, quizzes, and project prompts.
Course Map
Move from measuring uncertainty to building compression, channel, and inference systems.
Visual Models
Use these diagrams to keep formulas connected to compression, communication, and inference decisions.
Noisy-channel sketch
Use for mutual information, capacity, error correction, and decoder design.
Inference bottleneck
Use for model evaluation, leakage audits, and representation-learning questions.
How the labs work. The browser labs are small JavaScript instruments: entropy calculators, channel-capacity sliders, Huffman builders, and Bayes updates. They are deliberately transparent so the formulas stay visible.