Hacker News new | past | comments | ask | show | jobs | submit login
Artificial Intelligence: A Modern Approach (cs.berkeley.edu)
42 points by kqr2 on April 18, 2009 | hide | past | favorite | 28 comments



A classic but I'm not quite sure why it was posted? Initially thought they'd made the text available online, which would be cool. But alas.

Is the Google Code or design-a-cover thing the actual news here?


"A classic but I'm not quite sure why it was posted?"

and why is it getting so many upvotes? Extending this approach, if anyone is looking for free karma, just post separate links to all the classic CS books, without any comments or added explanation (3 separate submissions for the TAOCP Knuth books ;-)). Lots of karma for the taking ;-)

now that this is on the front page, here is something interesting. There's a 1000 $ prize for designing a cover for the third edition (see top right hand corner of teh page). The cover of the first two editions are very dense with references to AI history. If I could draw worth a damn I would have taken a shot at it.


Serious tease. I expected either the text available online, lecture videos or at the very least a courseware podcast from Berkeley. I guess I'll go back to staring at it at the bookstore...


Sign up for Gigapedia.org and you can get it. Email me if you have questions about how to DL it.


Agreed, although the code examples in python ( http://code.google.com/p/aima-python/ ) and lisp ( http://aima.cs.berkeley.edu/lisp/doc/overview.html ) are worth pointing out.


Does AI + Python have any practical use in production?

Most of the algorithms I've tested seem pretty slow. Although the linked examples are for learning.

From my research AI seems to be primarily LISP in academia and C in production.


I actually use a blend of C & Python, C for doing the heavy lifting and Python for the rapid prototyping.

Most AI algorithms have clearly defined segments that are very CPU intensive, by only using C in those areas I get the optimal balance of development speed and execution speed.


Ever since I learnt Cython, this has been my preferred combination as well. I can't overstate enough how great it is to be able to get the rapid prototyping of Python combined with being able to offload the heavy stuff to C.


Increasingly these days academic AI is done in Matlab or similar systems.


While the book is good, it lacks the statistical approach to AI and ML. Currently, most of the AI or ML is done in statistical fashion and I think this book does not do justice in introducing those topics well.


? chater 5 to 7 seem to cover that pretty well.

Part V Uncertain Knowledge and Reasoning 13 Uncertainty 14 Probabilistic Reasoning 15 Probabilistic Reasoning Over Time 16 Making Simple Decisions 17 Making Complex Decisions Part VI Learning 18 Learning from Observations 19 Knowledge in Learning 20 Statistical Learning Methods (pdf) 21 Reinforcement Learning Part VII Communicating, Perceiving, and Acting 22 Communication 23 Probabilistic Language Processing 24 Perception 25 Robotics Part VIII Conclusions 26 Philosophical Foundations 27 AI: Present and Future Bibliography (pdf and counts) Index (html or pdf)


what would you recommend to supplement Russel&Norvig?


"what would you recommend to supplement Russel&Norvig?"

Elements Of Statistical learning( http://www-stat.stanford.edu/~tibs/ElemStatLearn/ )


I would rather recommend Machine Learning video course from Stanford by Andrew Ng.

Pattern Recognition and Machine Learning by Christopher M. Bishop is good too


Everybody I meet in industry and academia has a copy of Bishop's "Pattern Recognition and Machine Learning". It's pretty comprehensive.

The most recent version of Russell and Norvig does have a solid statistics section in it, but they don't start it until the middle of the book.


My vote goes to Programming Collective Intelligence (http://oreilly.com/catalog/9780596529321/)


It appears a 3rd edition is on the way. I wonder what changes it will bring?

http://media.pearsoncmg.com/ph/esm/ecs_ai/


I like the AI Game Programming Wisdom Series (http://www.aiwisdom.com/).

Or maybe if we are posting up classics we could mention The Society of Mind by Marvin Minsky.


My name is Eliezer Yudkowsky, and I think this book is awesome.


I love this book as well, and I make sure to read it over regularly to spark different ideas - but it's expensive. Buying it was a bitterweet experience, to say the least.

Owell, it's at least the best you're likely to find.


My name is vang3lis and I'm an alcoholic. So what?


I know Elizier's name from Novamente, which is no small achievement.


Don't get me wrong, I like Elizier's posts on Overcoming Bias, but I disapprove of name calling when doing one sentence book reviews


Ah! In that case...

"Denny Crane!"


I want to clarify that this is a reference to the TV show Boston Legal, where William Shatner's named partner in a law firm, says his own name frequently, as if to say 'I'm Denny Crane and I'm the best'.


It bored me to tears. The "agent" orientation of the book does not address my needs for AI: data mining and inference. Norvig's companion Paradigms of Artificial Intelligence was a bliss however :-)


The ToC doesn't contain anything about the situated cognition, enactive cognition schools of thought, so I find applying the term AI to this book too broad. The topics covered seems to broadly fall under "automatic problem solving strategies" or something like that. Check out Rodney Brooks' robotics work for anything that feels like intelligence.


I sold my copy on Amazon when I realised I'd get all my money back.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: