Machine Learning

interactive book

Machine Learning

A seven-phase interactive book from the learning loop to generalization, optimization, deep networks, foundation models, agents, alignment, and evals. A first-year student should stay oriented; a graduate student should see the real machinery and the places where the field is still unsettled.

22
lessons
7
phases
21
demos
Syllabus → · Study guide → · Projects → · Glossary → · Start at lesson 1 →
Phase 1

The Learning Loop

What machine learning is, what it optimizes, and why prediction is a broader idea than forecasting.

You can describe any supervised learning system using data, model, loss, optimizer, and generalization.

  1. 🧠 lesson 1 The Prediction Machine What AI Actually Does (and Doesn't Do)
  2. 📈 lesson 2 The Learning Problem Fitting Curves to Chaos
Phase 2

Generalization and Measurement

How to know whether a model learned a pattern, memorized a sample, leaked data, or merely optimized the wrong score.

You can design train/validation/test splits, choose metrics, and explain why held-out performance matters.

  1. 🧪 lesson 3 The Generalization Gap Why Training Accuracy Is Not the Prize
  2. 📏 lesson 4 Measuring Models Accuracy Is Only One Lens
Phase 3

Linear Models and Optimization

Boundaries, weighted votes, gradients, and the first models simple enough to understand completely.

You can derive how a line becomes a classifier and how a local update changes the decision boundary.

  1. ✂️ lesson 5 Drawing Boundaries When Lines Become Decisions
  2. ⛰️ lesson 6 The Gradient Descent Lab Optimization as Controlled Motion
  3. ⚡ lesson 7 The Artificial Neuron Biology Inspires Math
Phase 4

Deep Networks

Layers, nonlinear representations, backpropagation, initialization, optimizers, and regularization.

You can explain how a scalar loss teaches many layers and why training stability is an engineering problem.

  1. 🏗️ lesson 8 Layers of Abstraction From Pixels to Concepts
  2. 🔄 lesson 9 The Credit Assignment Problem How Networks Learn From Mistakes
  3. 🛠️ lesson 10 Optimization in Practice Regularization, Stability, and the Training Recipe
Phase 5

Representations Across Space and Time

Architectures that exploit structure: images have locality, sequences have order, and attention creates content-addressed retrieval.

You can recognize the inductive bias behind CNNs, RNNs, and attention instead of memorizing architectures.

  1. 🔍 lesson 11 The Pattern Scanner Spatial Structure in Data
  2. 🔁 lesson 12 Memory in Networks Learning from Sequences
  3. 🎯 lesson 13 Focus, Not Memory The Revolution That Changed AI
Phase 6

Foundation Models

Transformers, scaling laws, generation, multimodal embeddings, retrieval, and tool-using systems.

You can trace how next-token training becomes a general interface for language, images, retrieval, and tools.

  1. 🏛️ lesson 14 The Architecture of Intelligence How GPT and Friends Actually Work
  2. 📊 lesson 15 The Scaling Hypothesis Why Bigger Models Keep Getting Better
  3. ✨ lesson 16 Creating, Not Classifying From Prediction to Generation
  4. 🔗 lesson 17 Connecting Senses When AI Sees, Hears, and Speaks
  5. 🧭 lesson 18 Retrieval, Tools, and Agents When a Model Is Part of a System
Phase 7

Agents, Alignment, and Evaluation

Learning from reward, aligning behavior with human intent, evaluating deployed systems, and thinking about frontier risk.

You can reason about policies, preference models, eval suites, reward hacking, and agent reliability.

  1. 🎯 lesson 19 Learning from Experience Trial, Error, and Reward
  2. 🤝 lesson 20 Aligning AI with Humans Teaching Preferences, Not Just Tasks
  3. 🧾 lesson 21 Evaluating AI Systems From Benchmarks to Behavioral Evidence
  4. 🛡️ lesson 22 The Road Ahead Challenges, Risks, and Possibilities