Resources
A small reference shelf — use the deck first
Do not try to read everything below. The lecture decks tell you what matters in a given week; these resources are dependable places to go deeper when you need another explanation, a derivation, or implementation context.
Core references
| Use this for | Resource |
|---|---|
| One broad, modern textbook | Simon J. D. Prince, Understanding Deep Learning |
| A rigorous second treatment | Christopher Bishop & Hugh Bishop, Deep Learning: Foundations and Concepts |
| Hands-on PyTorch | Dive into Deep Learning |
| Classical reference | Goodfellow, Bengio & Courville, Deep Learning |
Video companions
- Build from scratch: Andrej Karpathy’s Zero to Hero is ideal for the early implementation thread and language-model construction.
- Vision: CS231n is the most useful companion for CNNs, detection, and segmentation.
- Language and Transformers: CS224N provides a deeper treatment of sequence models and attention.
- Practical model work: the Hugging Face course is the best bridge to modern tooling.
Papers: read selectively
Read the original paper only after you understand the deck’s mechanism. The recurring landmarks are: ResNet, Attention Is All You Need, BERT/GPT, SimCLR/MAE/BYOL, VAE/GAN/DDPM, and FlashAttention. Individual deck endings name the relevant paper when it earns a close read.
Interactives
Interactive Articles hosts optional browser explainers for mechanics such as convolution, optimization, attention, contrastive learning, and diffusion. Follow the links on [I] lecture slides; they are the curated entry points rather than a separate parallel syllabus.
For each lecture
- Read the deck’s opening and its worked mechanism.
- Use one notebook or interactive only if it resolves a concrete question.
- Return to the relevant book or paper afterward, with the structure already in mind.