# Learning garden: book research and link verification

Checked on **2026-10-09 UTC**. Catalogue: **18 books or distinct volumes**, comprising **14 PDF resources** and **4 online books**. Six PDF entries also provide notebooks through their official source page. Descriptions are original summaries, not copied publisher blurbs. Levels are editorial guidance, not prerequisites stated by every author.

## Selection and access

The catalogue uses author websites, university pages, official project repositories, and author-submitted arXiv manuscripts. It links out to the authorized resource; it does not redistribute book files, cover art, or text. Free access does not imply an open redistribution license. Where the freely available edition is a draft or a notebook edition, the description says so. Publication years are distinguished from draft years.

HTTP checks used Python `urllib.request` with a browser user agent and a `Range: bytes=0-1023` request. A `206` status is the expected partial-content response. PDF checks read the returned bytes and checked for `%PDF-`, rather than inferring file type from a suffix. Sources were also inspected with web browsing for authorship, scope, publication details, and permission. PDF verification is a link health check, not a full editorial review of each book.

## Verified catalogue

| Book | Official provenance and access evidence | Resource check on 2026-10-09 UTC |
| --- | --- | --- |
| Dive into Deep Learning | [Official project](https://d2l.ai/) supplies the book, runnable notebooks, author list, and 2023 citation. [PDF](https://d2l.ai/d2l-en.pdf). | 07:46:56, HTTP 206, application/pdf, `%PDF-1.5`. |
| Understanding Deep Learning | [Author project](https://udlbook.github.io/udlbook/) and [official repository](https://github.com/udlbook/udlbook). Its live application and GitHub release API both identify v5.0.3, `UnderstandingDeepLearning_02_09_26_C.pdf`. The book was published in 2023; the linked file is an updated author edition. [PDF](https://github.com/udlbook/udlbook/releases/download/v5.0.3/UnderstandingDeepLearning_02_09_26_C.pdf). | 07:47:18, HTTP 206 after GitHub asset redirect, application/octet-stream, `%PDF-1.5`. |
| The Little Book of Deep Learning | [Author page](https://fleuret.org/francois/lbdl.html) identifies v1.3, updated May 12, 2026, and offers a free phone-formatted PDF under a noncommercial Creative Commons license. [PDF](https://fleuret.org/public/lbdl.pdf). | 07:46:57, HTTP 206, application/pdf, `%PDF-1.5`. |
| Neural Networks and Deep Learning | [Author reader](http://neuralnetworksanddeeplearning.com/) identifies Michael Nielsen, the 2015 citation, free online access, and CC BY-NC 3.0. The content is a foundational text, not a current framework API reference. | 07:47:49, HTTP 206, text/html. HTTPS has a hostname-mismatched certificate at verification; use the working author HTTP site. |
| Deep Learning | [Official companion](https://www.deeplearningbook.org/) identifies Goodfellow, Bengio, Courville, and the 2016 publication. Its FAQ explicitly states that a distributable PDF is not authorized. The complete HTML book is free. | 07:46:56, HTTP 206, text/html. Intentionally labelled Online book. |
| Deep Learning: Foundations and Concepts | [Official author site](https://www.bishopbook.com/) identifies Christopher M. Bishop and Hugh Bishop and embeds a free online reader. The site distinguishes that reader from the paid Springer PDF. The 2024 edition is listed in the catalogue. | 07:46:57, HTTP 206, text/html. Intentionally labelled Online book. |
| Deep Learning for Coders | [Official fast.ai repository](https://github.com/fastai/fastbook) contains the 2020 Howard/Gugger book as notebooks and links all chapters in Colab. Read the repository's prose restrictions separately from its GPL code license. | 07:46:57, HTTP 206, text/html. The catalogue explains the notebook edition and does not claim a free publisher PDF. |
| Mathematics for Machine Learning | [Authors' project](https://mml-book.github.io/) identifies Deisenroth, Faisal, Ong, the April 2020 publication, and a commitment to keep updated PDFs free. Includes notebook tutorials. [PDF](https://mml-book.github.io/book/mml-book.pdf). | 07:46:57, HTTP 206, application/pdf, `%PDF-1.5`. |
| An Introduction to Statistical Learning with Python | [Official book site](https://www.statlearning.com/) identifies all five authors, the 2023 Python edition, labs, and a download link. Its [author download page](https://hastie.su.domains/ISLP/ISLP_website.pdf.download.html) redirects to [this official Drive PDF](https://drive.google.com/uc?export=download&id=1ajFkHO6zjrdGNqhqW1jKBZdiNGh_8YQ1). | 07:47:56, HTTP 206 after redirect to drive.usercontent.google.com, application/octet-stream, `%PDF-1.3`. Drive viewer separately returns HTTP 200 with title ISLP_website.pdf. |
| The Elements of Statistical Learning | [Author site](https://hastie.su.domains/ElemStatLearn/main.html) identifies the 2009 second edition. Its [download page](https://hastie.su.domains/ElemStatLearn/download.html) states publisher permission and offers the corrected twelfth printing from January 2017. The linked intermediary redirects to [this official Drive PDF](https://drive.google.com/uc?export=download&id=1--Wo5Hcl2y_v3DL-tGgcJHRPdtjiVYjS). | 07:47:59, HTTP 206 after redirect to drive.usercontent.google.com, application/octet-stream, `%PDF-1.6`. Drive viewer separately returns HTTP 200 with title ESLII_print12_toc.pdf. |
| Probabilistic Machine Learning: An Introduction | [Author companion](https://probml.github.io/pml-book/book1.html) identifies Kevin P. Murphy, the 2022 publication, CC BY-NC-ND author draft, and notebooks. Its own link selects [latest official PDF](https://github.com/probml/pml-book/releases/latest/download/book1.pdf). | 07:47:18, HTTP 206 after GitHub asset redirect, application/octet-stream, `%PDF-1.7`. |
| Probabilistic Machine Learning: Advanced Topics | [Author companion](https://probml.github.io/pml-book/book2.html) identifies Kevin P. Murphy, the 2023 publication, CC BY-NC-ND author draft, and notebooks. Its own link selects [latest official PDF](https://github.com/probml/pml2-book/releases/latest/download/book2.pdf). | 07:47:19, HTTP 206 after GitHub asset redirect, application/octet-stream, `%PDF-1.7`. |
| Speech and Language Processing | [Stanford author page](https://web.stanford.edu/~jurafsky/slp3/) identifies Jurafsky and Martin and the August 19, 2026 third-edition draft. It links [this complete manuscript PDF](https://web.stanford.edu/~jurafsky/slp3/ed3book_aug26.pdf). This is an evolving draft; some planned chapters remain incomplete. | 07:46:58, HTTP 206, application/pdf, `%PDF-1.7`. |
| Graph Representation Learning | [McGill author page](https://www.cs.mcgill.ca/~wlh/grl_book/) explicitly states publisher permission to host the pre-publication draft. The [PDF](https://www.cs.mcgill.ca/~wlh/grl_book/files/GRL_Book.pdf) identifies Hamilton and 2020. | 07:46:59, HTTP 206, application/pdf, `%PDF-1.5`. |
| Reinforcement Learning: An Introduction | [Sutton's author page](http://incompleteideas.net/book/the-book-2nd.html) links the [second-edition PDF](http://incompleteideas.net/book/RLbook2020.pdf), code, errata, and solutions. The book is the 2018 second edition; the filename reflects the corrected file. | 07:46:58, HTTP 206, application/pdf, `%PDF-1.3`. HTTPS presents a self-signed certificate at verification; the working author HTTP link is retained. |
| Introduction to Machine Learning Systems | [Official Volume I](https://mlsysbook.ai/vol1/) identifies Vijay Janapa Reddi as author, editor, and curator, Harvard affiliation, 2026 publication, and a directly downloadable [PDF](https://mlsysbook.ai/vol1/assets/downloads/Machine-Learning-Systems-Vol1.pdf). [Project repository](https://github.com/harvard-edge/cs249r_book). | 07:46:59, HTTP 206, application/pdf, `%PDF-1.4`. |
| Machine Learning Systems at Scale | [Official Volume II](https://mlsysbook.ai/vol2/) identifies Vijay Janapa Reddi and the same project. Its online version was updated in 2026, while the landing page lists a future 2027 publisher edition. The catalogue therefore says 2026 draft. [PDF](https://mlsysbook.ai/vol2/assets/downloads/Machine-Learning-Systems-Vol2.pdf). | 07:47:00, HTTP 206, application/pdf, `%PDF-1.4`. |
| The Principles of Deep Learning Theory | [Author arXiv record](https://arxiv.org/abs/2106.10165) identifies Roberts, Yaida, Hanin, and the August 2021 preprint revision, with a 2022 published-book reference. [PDF](https://arxiv.org/pdf/2106.10165). The catalogue specifies 2021 preprint. | Web tool successfully opened and parsed application/pdf, 471 pages, on 2026-10-09. Local Python certificate validation failed for arxiv.org, so no local byte-signature success is claimed. |

## Important implementation details

- Do not replace the current ISLP or ESL URLs with their former `.pdf` paths at hastie.su.domains. Both former direct paths returned HTTP 404. The author's current links intentionally route to Google Drive.
- GitHub and Google Drive may deliver PDF bytes with `application/octet-stream` or attachment headers. A normal link opens the resource in a new tab, but the user's browser and the origin server ultimately determine whether it displays or downloads. Do not promise an embedded viewer for those links.
- The catalogue is curated data fetched during research, not a live client-side scraper. This avoids fragile CORS dependencies and rate-limited GitHub requests during page load. The exported `booksCheckedAt` value records when these links were checked.
- `PDF + notebooks` means the main action opens the PDF while the accompanying source link leads to the author's page with notebooks. It is not a bundled download.
- No third-party scans, paid-only editions, sample-only PDFs, or unlicensed mirrors were included. Goodfellow and Bishop remain authorized online reader links. The fast.ai entry is the author's notebook edition.
- Classic books remain useful for fundamentals, but examples may rely on old software versions. The catalogue describes conceptual scope rather than claiming every historical code example works with present-day packages.

## Research breadth

The search covered core deep learning, practical implementation, mathematics, statistical learning, probabilistic inference, modern NLP and retrieval, graph learning, reinforcement learning, systems engineering, and theory. Primary materials were opened across D2L, official GitHub projects, the authors' independent sites, Stanford, McGill, Harvard's ML Systems project, and arXiv. Other search results were used for discovery only, never as the source of book files.