---
title: "Deep Learning"
authors:
  - "Ian Goodfellow"
  - "Yoshua Bengio"
  - "Aaron Courville"
year: 2016
venue: "MIT Press"
arxiv: null
doi: "10.7551/mitpress/7137.001.0001"
source: "https://www.deeplearningbook.org/"
topics:
  - "deep-learning"
  - "textbook"
seed_rank: 961
seed_batch: "user-add-2026-08-13"
reviewed: "2026-08-13"
pool: "languages"
relevance_score: 8
lineage: categorical-deep-learning
cites:
  - title: "ImageNet classification with deep convolutional neural networks"
    url: "https://doi.org/10.1145/3065386"
    year: 2017
    doi: "10.1145/3065386"
  - title: "Deep Residual Learning for Image Recognition"
    url: "https://arxiv.org/abs/1512.03385"
    year: 2015
    arxiv: "1512.03385"
see:
  - "949-imagenet-classification-with-deep-convolutional-neural-nets"
  - "947-deep-residual-learning-for-image-recognition"
---

# Deep Learning

## One-sentence takeaway

Goodfellow, Bengio, and Courville's 2016 MIT Press textbook is the standard informal account of pre-Transformer deep learning — MLPs, convnets, sequence models, and backprop — that Weaves cites as the baseline its algebra is meant to replace.

## Why it matters here

Background literacy, not a Broadside-specific result. Kept because the Weaves bibliography treats the book as a first-class reference for the informal broadcasting and layer-composition story that NCD/Weaves then make compositional.

## Key ideas

- Free HTML at https://www.deeplearningbook.org/; MIT Press DOI `10.7551/mitpress/7137.001.0001`.
- Three parts: applied-math background, modern practice (feedforward nets, conv, recurrence, regularization), and research topics (autoencoders, representation learning, structured probabilistic models).
- The architecture cards (AlexNet / ResNet / Transformer) carry the specifics; this card is the textbook that still writes broadcasting in pseudocode rather than as a morphism.

## Caveats

## Links

- https://www.deeplearningbook.org/
- DOI: https://doi.org/10.7551/mitpress/7137.001.0001