Resources for Getting Started with Deep Learning

AI@MPCDF
1

Hi everyone,

I’m often asked for good starting points to learn about deep learning, so I thought I’d gather some recommendations here.

Feel free to add any books, papers, tutorials, or links that you think are helpful for anyone beginning their deep‑learning journey!

4

The classic textbook “Deep Learning” by Goodfellow, Bengio, and Courville offers a solid theoretical foundation, but its treatment can feel a bit dense for newcomers. In my experience, the most engaging and hands‑on introduction is “Hands‑On Machine Learning with Scikit‑Learn, Keras & TensorFlow” by Aurélien Géron. That said, I wouldn’t recommend starting there today (especially the deep‑learning chapters) since the material is tightly tied to TensorFlow/Keras, which may not be the most beginner‑friendly entry point any longer.

1 Like
5

I’d also recommend Bishop’s Pattern Recognition and Machine Learning for theoretical foundation.

For practical deep learning with coding examples, check out the UvA Deep Learning Course notebooks: Welcome to the UvA Deep Learning Tutorials! — UvA DL Notebooks v1.2 documentation.

For a more modern, hands-on perspective on generative models, see Tomczak’s Deep Generative Modelling (https://jmtomczak.github.io/dgm_book.html).

1 Like
6

Not a book, but a very cool animation that explains the back‑propagation mechanism can be found here:

2 Likes
7

I am a huge fan of Andrej Karpathy’s Neuronal Network: Zero to Hero YouTube Playlist:
https://www.youtube.com/playlist?list=PLAqhIrjkxbuWI23v9cThsA9GvCAUhRvKZ

It covers the basic details from a minimalist auto-grad implementation up to usable GPT-2-style transformers without the bells and whistles of production-ready frameworks or the heavy mathematical burden of more academic introductions into Deep Learning. A truly great resource for someone with some Python knowledge to start with!

In my opinion, Andrej’s experience as principal instructor for the Stanford class CS 231n: Convolutional Neural Networks for Visual Recognition really shines through.

Best wishes,
Eric

8

This is very nice book for Diffusion and Score-matching models: https://arxiv.org/abs/2403.18103