Hacker News Reader

Comment by arihant | Hacker News Reader

You cannot really go practical in AI without academic rigor. You can do recipes of what's already been done using a TensorFlow book, but that's how far one can go. If one is serious in getting in AI today, a great way to do is read the following books, in order:

1. AI: A Modern Approach by Stuart Russell and Peter Norvig.

2. Deep Learning by Ian Goodfellow and Yoshua Bengio.

It is amazing how approachable both books are for beginners, but you will be diving a lot into academic stuff as you go along.

View on HN

I love Paradigms of Artificial Intelligence by Peter Norvig much more than AIMA, which I found excessively encyclopedic and shallow.

While some will argue it is dated, I think it presents many timeless ideas that will get in vogue soon with little tweaks to their inference schemes.

Same for The Art of Prolog.

If you're shipping anything after 2010, you're not going to get within an order of magnitude within state of the art with that book (PAI), unfortunately.

There's basically no numerics in that book about anything that'll past muster at NIPS or ICML nowadays or would be shipped by one of the big corporate AI labs, I'm sorry to say.

Art of Prolog was over of the best books I've read. As Prolog itself is one of my favorite languages. I wish there were more of both.

CTM by Van Roy & Haridi?

I will check it out. Thanks.

I had the feeling it was more of a Common Lisp book than AI book.

It's both, but the AI coversge is quite out of date.

PAIP is one of my favorite books ever, but taken as a book about the craft of programming, not about AI. AI has grown, and the broadness of AIMA matches the subject. (It does need another update, and I heard they're working on one.)

PAIP is also very, very high on my list as well. I'm very pro-lisp and still develop new projects in lisp (CL, Scheme) and promote it when/where I can. But the person who picks up PAIP wanting to learn AI might not necessarily want to worry about picking up lisp programming skills at the same time (nor is learning lisp strictly necessary). This is why AIMA is still the best option, IMO, because it employs a language agnostic approach.

It's Paradigms of Artificial Intelligence Programming (PAIP) http://norvig.com/paip.html

AIMA provides better introduction for wider area of subjects but PAIP is one of most elegant and timeless books for both programming and old school AI.

Why not both?

PAIP is awesome. Will make you a better thinker and developers. A truly hidden gem.

Sutton & Barto's Reinforment Learning complete this triumvirate

https://webdocs.cs.ualberta.ca/~sutton/book/the-book-2nd.htm...

David Silver's Reinforcement Course is based on Sutton & Barto

https://www.youtube.com/watch?v=2pWv7GOvuf0

I liked all three of them. I also feel Murphy's MLAPP provides additional useful material and is well written.

For whatever reason, Sutton's was the first serious book I read in any area of AI. The balance between explaining the history, the concepts and the code is handled really well.

I have both, and I agree these are great. I really appreciate the style of the Goodfellow book, it's very approachable.

Reply on news.ycombinator.com