Further Reading: Chapter 36

The Road to Advanced: Deep Learning, Causal Inference, MLOps, and Where to Go Next

Deep Learning Fundamentals

1. Deep Learning --- Ian Goodfellow, Yoshua Bengio, and Aaron Courville (2016) The canonical deep learning textbook. Part I covers the mathematical foundations (linear algebra, probability, numerical optimization). Part II covers the core architectures (feedforward networks, CNNs, RNNs, autoencoders). Part III covers research frontiers. Dense and rigorous --- not a first read for most practitioners, but the definitive reference. MIT Press. Free online at deeplearningbook.org.

2. Dive into Deep Learning --- Aston Zhang, Zachary Lipton, Mu Li, and Alexander Smola (2023) An interactive, code-first deep learning textbook with implementations in PyTorch, TensorFlow, and JAX. Every concept is accompanied by runnable code. More accessible than Goodfellow et al. and more rigorous than a blog post. Covers CNNs, RNNs, transformers, and attention. Free online at d2l.ai.

3. fast.ai "Practical Deep Learning for Coders" --- Jeremy Howard and Sylvain Gugger (free course) The most effective introduction to deep learning for practitioners. The course follows a top-down approach: you train state-of-the-art models on day one, then progressively learn how they work. Covers vision, NLP, tabular data, and recommendation systems. The accompanying textbook (Deep Learning for Coders with fastai and PyTorch, O'Reilly 2020) is excellent. course.fast.ai.

4. "Why do tree-based models still outperform deep learning on typical tabular data?" --- Grinsztajn, Oyallon, and Varoquaux (NeurIPS 2022) The empirical paper behind this chapter's claim that gradient boosting beats deep learning on tabular data. The authors benchmark across 45 datasets and find that tree-based models (random forests, gradient boosting) consistently outperform deep learning (MLP, ResNet, FT-Transformer) on medium-sized tabular datasets. The gap is largest when features have irregular patterns and when uninformative features are present. Available on arXiv (2207.08815).

PyTorch and TensorFlow

5. Programming PyTorch for Deep Learning --- Ian Pointer (2019) A practical guide to PyTorch for developers coming from scikit-learn. Covers tensors, autograd, building custom models, transfer learning, and deployment. The early chapters bridge the gap between scikit-learn's .fit() API and PyTorch's explicit training loops. O'Reilly.

6. PyTorch Official Tutorials (pytorch.org/tutorials) The official tutorials are surprisingly well-written. Start with "Learning PyTorch with Examples" (builds a network from numpy to autograd to nn.Module) and "Transfer Learning Tutorial" (fine-tuning ResNet). Free.

7. Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow --- Aurelien Geron (3rd edition, 2022) The most popular ML textbook for practitioners. Part II covers deep learning with TensorFlow/Keras. Geron's writing is clear and practical, with production-oriented code. If you prefer TensorFlow over PyTorch, this is your starting point. O'Reilly.

Causal Inference

8. Causal Inference: The Mixtape --- Scott Cunningham (2021) The most accessible introduction to causal inference for data scientists. Covers potential outcomes, matching, difference-in-differences, regression discontinuity, instrumental variables, and synthetic control. Written in a conversational style with code examples in Stata, R, and Python. Free online at mixtape.scunning.com. Yale University Press.

9. The Effect: An Introduction to Research Design and Causality --- Nick Huntington-Klein (2021) A modern causal inference textbook with outstanding visualizations. Huntington-Klein explains directed acyclic graphs (DAGs), identification strategies, and the logic of causal reasoning with unusual clarity. Lighter on mathematical formalism than Cunningham, heavier on intuition. Free online at theeffectbook.net. Chapman and Hall/CRC.

10. Causal Inference for The Brave and True --- Matheus Facure Alves (2022) A Python-focused causal inference textbook. Each chapter is a Jupyter notebook with runnable code. Covers A/B testing, propensity scores, difference-in-differences, regression discontinuity, and instrumental variables. Excellent bridge between the theory in Cunningham/Huntington-Klein and practical implementation. Free online at matheusfacure.github.io/python-causality-handbook.

11. The Book of Why: The New Science of Cause and Effect --- Judea Pearl and Dana Mackenzie (2018) Pearl's popular-audience introduction to causal reasoning. Covers the ladder of causation (association, intervention, counterfactual), DAGs, do-calculus, and the history of causal inference. Not a technical reference, but essential for understanding why causal inference is a fundamentally different enterprise from prediction. Basic Books.

12. "Introduction to Causal Inference" --- Brady Neal (free course, 2020) A video lecture series covering causal inference from first principles. Covers potential outcomes, DAGs, backdoor criterion, do-calculus, difference-in-differences, instrumental variables, and more. Clear, rigorous, and free. Available at bradyneal.com/causal-inference-course.

MLOps and Production ML

13. Designing Machine Learning Systems --- Chip Huyen (2022) The definitive MLOps textbook. Covers data engineering, feature engineering, model development, deployment, monitoring, and infrastructure. Huyen draws on her experience at NVIDIA, Snorkel AI, and as a Stanford instructor. Chapters 7 (Model Deployment) and 8 (Data Distribution Shifts) are particularly strong. O'Reilly.

14. "MLOps: Continuous delivery and automation pipelines in machine learning" --- Google Cloud (2020) The whitepaper that defined the MLOps maturity levels (0, 1, 2) used in this chapter. Concise, practical, and freely available. Read this before diving into any MLOps tool. cloud.google.com/architecture/mlops-continuous-delivery-and-automation-pipelines-in-machine-learning.

15. Machine Learning Engineering --- Andriy Burkov (2020) A practical guide to building ML systems in production. Covers project scoping, data collection, feature engineering, model training, serving, and monitoring. Written by a practitioner for practitioners. Shorter and more focused than Huyen, with strong coverage of the non-modeling aspects of ML engineering. True Positive Inc.

16. Made With ML --- Goku Mohandas (free, madewithml.com) A comprehensive, free MLOps course covering the full ML engineering lifecycle: data, modeling, developing, serving, and testing. Hands-on with modern tools (Ray, MLflow, FastAPI). Excellent for practitioners who learn by building.

NLP and Large Language Models

17. Hugging Face NLP Course (huggingface.co/learn/nlp-course) The best free introduction to modern NLP. Covers the Hugging Face ecosystem (Transformers, Datasets, Tokenizers), fine-tuning BERT, text classification, named entity recognition, question answering, and summarization. Hands-on from the first lesson. Free.

18. Speech and Language Processing --- Dan Jurafsky and James Martin (3rd edition, in progress) The NLP textbook. Covers everything from regular expressions to transformers. The third edition (free online at web.stanford.edu/~jurafsky/slp3/) is continuously updated. Dense but comprehensive. Use it as a reference alongside the Hugging Face course.

19. "Attention Is All You Need" --- Vaswani et al. (NeurIPS 2017) The paper that introduced the transformer architecture. Understanding this paper (even at a high level) is a prerequisite for working with modern NLP. The self-attention mechanism, multi-head attention, and positional encoding are now foundational concepts. Available on arXiv (1706.03762).

Computer Vision

20. CS231n: Convolutional Neural Networks for Visual Recognition --- Stanford (free lecture notes) The canonical computer vision course. Lecture notes cover image classification, CNNs, object detection, segmentation, and generative models. The 2017 lecture videos (available on YouTube) remain relevant. cs231n.github.io.

21. Deep Learning for Vision Systems --- Mohamed Elgendy (2020) A practical introduction to computer vision with deep learning. Covers CNNs, transfer learning, object detection (YOLO, SSD), and image segmentation. Lighter on theory than CS231n, heavier on implementation. Manning.

Experimentation and A/B Testing (Advanced)

22. Trustworthy Online Controlled Experiments --- Kohavi, Tang, and Xu (2020) The definitive reference on A/B testing at scale. Covers statistical foundations, sample size calculation, network effects, long-term effects, and the organizational challenges of experimentation. Written by the team that built Microsoft's experimentation platform. Cambridge University Press.

23. "Causal Inference and Discovery in Python" --- Aleksander Molak (2023) A hands-on guide to implementing causal inference methods in Python. Covers graphical models, do-calculus, propensity scores, difference-in-differences, and causal discovery algorithms. Fills the gap between causal inference theory and Python implementation. Packt.

How to Use This List

If you are starting with deep learning, begin with fast.ai (item 3) for the practical introduction and Dive into Deep Learning (item 2) for the conceptual depth. Read Grinsztajn et al. (item 4) to calibrate your expectations about when deep learning actually helps.

If you are starting with causal inference, begin with Cunningham (item 8) or Huntington-Klein (item 9) for the theory, and Facure Alves (item 10) for the Python implementation. Read Pearl (item 11) for the philosophical foundation.

If you are starting with MLOps, begin with Huyen (item 13) for the comprehensive overview and the Google whitepaper (item 14) for the maturity framework. Use Made With ML (item 16) for hands-on practice.

If you are starting with NLP, begin with the Hugging Face NLP Course (item 17). It is the fastest path from "I know scikit-learn" to "I can fine-tune a transformer."

If you are starting with computer vision, begin with fast.ai (item 3, which starts with vision) and the CS231n lecture notes (item 20).

Regardless of your path, read Huyen (item 13). Every data scientist eventually needs to put models into production, and this book is the best preparation for that transition.

This reading list supports Chapter 36: The Road to Advanced. Return to the chapter to review the learning paths before choosing your resources.