Sampling of #deeplearning textbooks

Here are some other books about ML and Deep Learning that I recommend: shepherd.com/best-books/mac…

If you are enjoying my book and want to learn more about transformers and language models, you might want to check out this set of articles I wrote for Borealis AI.

Part II of our blog series shows that in the infinite width limit, neural networks become linear in their parameters. Surprisingly, this allows us to write closed-form solutions for the training evolution. These dynamics depend on the neural tangent kernel (NTK).

If you are having trouble getting hold of "Understanding Deep Learning", it's because the first printing has sold out. New copies should be available early March. You can pre-order via: penguinrandomhouse.com/books/730887/u… Full PDF still available via udlbook.com.

As of this morning, there have been 250,000 downloads of Understanding Deep Learning (udlbook.com). I'm amazed, humbled, and extremely grateful for everyone's support. It's still out of print, but will be available again in the next couple of weeks!

I'm delighted to say that "Understanding Deep Learning" (pdf at udlbook.com) has just been reprinted by The MIT Press @mitpress.bsky.social and is available once more from Amazon and other booksellers.

New fancy website for "Understanding Deep Learning" book at udlbook.com including links to many other articles I've written on NLP, responsible AI, graphical models, optimization, computer vision, math for ML, transformers etc.

Foundations of Computer Vision by Antonio Torralba, Phillip Isola, and Bill Freeman available today and published by The MIT Press @mitpress.bsky.social. I haven't read it yet, but almost certain to be worth reading given the pedigree of the authors.

A very positive review of Understanding Deep Learning (pdf at udlbook.com) from Ge Wang in Nature Machine Intelligence. Thank you so much for your kind words! nature.com/articles/s4225…

Here's a new blog I wrote for Borealis AI about the neural tangent kernel: tinyurl.com/39yf49mn This is a super-interesting area that is hard to understand and it's a notable omission from my book (udlbook.com). I've made it as clear as possible.

Really cool work from my former colleagues at Borealis. Reduces GPU memory in training / fine tuning by a factor of three. Faster models! Larger batches! And only one or two easy code changes to make this happen.

New blog for Borealis AI is 1st of 4 on Bayesian methods: borealisai.com/research-blogs… Suitable for those new to this area, and comes with working code. 3 more blogs to follow including stuff on Bayesian neural networks and NNGPs. Previous articles via: udlbook.com

Learning or teaching from "Understanding Deep Learning" (pdf at udlbook.com)? There are now interactive versions of 9 of the figures, covering 1D linear regression, shallow and deep networks, and optimization: udlbook.github.io/udlfigures/ (Not suitable for phones)

FYI Understanding deep learning is currently on sale, reduced by 18% at Amazon US: tinyurl.com/2xb6czt6 and by 31% at Amazon UK: tinyurl.com/y4prn3rx Bargain! Of course, the PDF is and will remain forever public at udlbook.com as well.

Blog 2 of 4 on Bayesian methods in ML for Borealis AI: borealisai.com/research-blogs… Suitable for those new to this area, and comes with working code. Two more to follow including stuff on Bayesian neural networks and NNGPs. Previous articles at: udlbook.com