Reference Textbooks
The canonical books for when the TL;DRs aren't enough. Many are free PDFs online.
General ML & deep learning
- Goodfellow, Bengio, Courville — Deep Learning The classic deep learning text. Free HTML version. Slightly dated on modern architectures (no transformers in depth) but the foundations chapters (6–8) are still the standard reference.
- Hastie, Tibshirani, Friedman — Elements of Statistical Learning The classical-ML reference. Free PDF. Dense but exhaustive — every section eventually gets cited somewhere.
- James, Witten, Hastie, Tibshirani — Introduction to Statistical Learning Friendlier little sibling of ESL. Free PDF, with Python and R versions. Best entry point for classical ML with statistical rigor.
- Murphy — Probabilistic Machine Learning Modern, comprehensive, free. Two volumes (foundations + advanced). Strongest treatment of probabilistic methods. Vol 1 is the default modern ML textbook.
- Bishop — Pattern Recognition and Machine Learning 2006 but still relevant. Beautiful Bayesian-flavoured treatment. Free PDF. The clearest source on EM, mixture models, and variational methods.
Specialty topics
- Sutton & Barto — Reinforcement Learning The standard RL textbook. Free PDF. Foundational for understanding policy gradients, value functions, and the modern derivatives.
- Jurafsky & Martin — Speech and Language Processing 3rd edition, in progress, free draft. The canonical NLP textbook. Covers everything from regex to transformers.
- Rasmussen & Williams — Gaussian Processes for ML Free PDF. The reference on GPs from regression to classification to sparse approximations.
- Molnar — Interpretable Machine Learning The reference for SHAP, LIME, partial dependence, counterfactual explanations. Free online book.
- Barocas, Hardt, Narayanan — Fairness and ML The standard text on fairness, accountability, and ethical considerations in ML.
Practitioner-oriented
- Huyen — Designing Machine Learning Systems Chip Huyen on the engineering of ML systems — data, deployment, monitoring, drift. The best single book on putting ML in production.
- Lakshmanan, Robinson, Munn — Machine Learning Design Patterns Patterns and anti-patterns from real production systems. Pairs well with Huyen.