Beginners AI

[El Corazon]

brianwelsch wrote:

I'm looking for concepts at this point rather than algorithms and in depth theory.

I would start on the net if I were you, save the books for which areas are of interest to you. You can spend your whole life (and some do, or lack thereof) dedicated to a subportion therein. http://en.wikipedia.org/wiki/Portal:Artificial_intelligence[^] State Engines -> Fuzzy Logic -> Neural Nets -> Hybrid Fuzzy sets State Engines are where we all started from, and many programming systems still are. You are either at a state, or not, you have an integer value or a floating point value, etc. The goal of a state engine is to define the operative states such that from beginning to end you define enough to get through operations. The good news is that state engines, being finite states, are easily repetative. The bad news is they cannot evaluate new inputs or partial states. A state engine declares you are in a room or not, but if you stand in the doorway, how is it defined? Fuzzy logic can evaluate partial states and more. :) The advantage from starting at state-engine defined logic is that it leads directly into Fuzzy Logic intuitively, Neural Nets and hybrid/adaptive/learning systems come as you learn more.... Now see what you did by getting me started talking?? :rolleyes: Good luck! _________________________ Asu no koto o ieba, tenjo de nezumi ga warau. Talk about things of tomorrow and the mice in the ceiling laugh. (Japanese Proverb)

[Ravi Bhavnani]

Brian, AI (like "medicine") is a very wide field encompassing pattern recognition, inference, heuristics, common sense knowledge, learning from experience to name a few. All these terms are taken from here[^]. My academic and professional background is in expert systems[^] so I'm partial to that area, and more recently common sense knowledge (see this[^] link for a nice intro). The April '85 issue of Byte Magazine[^] (now extinct) that focused on AI changed my life. Another book that impacted me greatly was Buchanan & Shortliffe's book on reasoning, specifically the chapters devoted to Mycin[^]. Both good reads, as are also introductory materials by Patrick Henry Winston[^], John McCarthy[^] and Marvin Minsky[^]. Hope this helps! /ravi My new year's resolution: 2048 x 1536 Home | Music |

[Graham Shanks]

According to Amazon[^] its been out since March 9. The pain is that I'm in the UK - US price from Amazon is $44.07, while UK price from Amazon.co.uk is £44.60 (about $74). I was in Alabama last week (the first of my biannual trips to the US) but the book store didn't have it (and I didn't realise that it was out yet). I hate the fact that computer books in particular are so much more expensive in the UK than in the US :mad: I guess that it is worth ordering from amazon.com and paying the international shipping (about $10) - after a bit more searching from other on-line retailers Graham

[brianwelsch]

Great! Thanks for the links. Learning from experience right now is the most intriguing to me, but that might be because I've started reading A User's Guide to the Brain[^], which initially deals with the development of neurons and perception. I also have How the Mind Works[^]. The opening of which discusses some issues with robots sensing their environment. BW

---

If you're not part of the solution, you're part of the precipitate.
-- Steven Wright

[Graham Nimbley]

Try Gamedev. It's always the first place I turn to for maths/physics/AI related subjects. There are a huge list of articles to wet your feet! http://www.gamedev.net/reference/list.asp?categoryid=18[^] Graham

[brianwelsch]

Jeffry J. Brickley wrote:

The advantage from starting at state-engine defined logic is that it leads directly into Fuzzy Logic intuitively, Neural Nets and hybrid/adaptive/learning systems come as you learn more....

I'll keep that mind, so as not to get ahead of myself and stop in a haze of confusion. :) Thanks for the link. BW

---

If you're not part of the solution, you're part of the precipitate.
-- Steven Wright

[brianwelsch]

Ed.Poore wrote:

I've found it's always helpful to start at Wikipedia, you never know where things end up

Good point, thanks. BW

---

If you're not part of the solution, you're part of the precipitate.
-- Steven Wright

[brianwelsch]

:cool: thanks. BW

---

If you're not part of the solution, you're part of the precipitate.
-- Steven Wright

[brianwelsch]

I started reading up on some basics of how the brain develops and how neurons form connections while learning new tasks, etc. Pretty neat. It helps understand why repetition increases long-term recollection. Those Gibson books actually sound like I might enjoy them too. I'll toss them onto my lengthening wish list at Amazon. BW

---

If you're not part of the solution, you're part of the precipitate.
-- Steven Wright

[El Corazon]

Graham Shanks wrote:

According to Amazon[^] its been out since March 9.

Oh well, no wonder.... :) I have been swamped since March 6th... It was here during compiles, work, and sleep and not much more. :) _________________________ Asu no koto o ieba, tenjo de nezumi ga warau. Talk about things of tomorrow and the mice in the ceiling laugh. (Japanese Proverb)

[El Corazon]

brianwelsch wrote:

I'll keep that mind, so as not to get ahead of myself and stop in a haze of confusion. Thanks for the link.

Just write your own Prolog and Lisp compilers, afterwards you will understand a lot more about inter-set dynamics and it will all be crystal clear.... remember what they say about asking an engineer what time it is? he'll tell you how to build a clock.... Well, ask a software engineer how to do something, he'll tell you how to write a compiler that does it for you... :laugh: _________________________ Asu no koto o ieba, tenjo de nezumi ga warau. Talk about things of tomorrow and the mice in the ceiling laugh. (Japanese Proverb)

[donnuno]

Hello Brian, I´m studying computer engineering, and last semester, i studyied AI. The best way for you to understand it, is to read a PROLOG beginers tutorial. Here you have a good one to start with http://www.coli.uni-saarland.de/~kris/learn-prolog-now/html/prolog-notes.pdf[^] if you want to go further, i advise to read this book, it´s a bit old, but very good explaining the concepts, Artificial Intelligence A Modern Approach Stuart J. Russell and Peter Norvig Good luck Nuno -- modified at 19:20 Wednesday 12th April, 2006

[KChandos]

If you're interested in delving into the Expert Systems arena (an AI close cousin), try http://www.ghg.net/clips/CLIPS.html Kinda cool.

[mbrezu2]

Hi, I think you should have a look at Lisp (or Scheme). A good starting point (yet rather steep) is Structure and Interpretation of Computer Programs by Abelson, Sussman & Sussman. It's the book used to teach the MIT 6001 course and it provides a good way to start thinking using "the Lisp way". Even if you don't ever program Lisp in real life, "the Lisp way" is becoming more and more influential (I hear C# 2 has closures (implemented by copying stack frames on the heap - I might have heard wrong here) :p, I know STL implements lots of functional stuff, I also know C# and Java are garbage collected etc. ). Another book (haven't read it yet, I mean to order it ASAP) is Peter Norvig's "Paradigms of Artificial Intelligence Programming". Anyway, after SICP, any programming book will read a little different ;) The point with Lisp/Scheme is that you stop worrying about low level implementation so much and have time to explore ideas. The programs won't run as fast, but, hey, in the beginning you'll be learning/prototyping anyway. Just a thought - hope this won't start a flame or anything, it wasn't meant to do that :p Cheers Miron

[Siiilon]

Well. Maybe I wont recommend you a particular literature, but I want to encourage you to study math for real understanding of AI. AI is very wide area and I am interested in its particular part. So according to what do you expect of AI you should lead particular topics. Since I am not great in math I am still able to do that. You dont need to be math genius, but without it it's almost impossible. Please, dont avoid AI's most powerfull weapon - "math". Math is for everyone and is easy if you dont fear of it. Regards, Michal

[Pimenov]

Hello Brian. For the start-point I could recommend you the Alice-bot project: www.alicebot.org[^] Of cause it is "far from" real AI :O) - it is just AIML ("AI markup language") - the links-list between input-words and output phrazes. But understanding this, you could "grasp", that the basement of ANY AI-programms should be "neural-networks" (like in human brain). On the same site I recommend you to go to the page with other "likewise" projects-list ("Bot indistry survay"): http://www.alicebot.org/aimlbots.html[^] If you are interested in "more detailed" discussion of practical applying (programming) of neural-networks (simplified approach to AI with "on the fly learning") - I could invite you to write to me personally: pvv@federalspace.ru (I am the "head" (:O) of the ROSCOSMOS (Russian NASA :O) AI-project). ( http://ai.federalspace.ru/ChatNETE.aspx) regards, Valeriy V Pimenov.

[Niclas Hartmann]

Hi brianwelsch, two textbook recommendations: overview, beginner level: "Artificial Intelligence - A Modern Approach" by Russel & Norvig. An almost complete AND detailed summary of the field (planning, learning, decision making, fuzzy logic, natural language processing, expert system, neuronal networks - you name it), including both theoretical (discrete math and mathematical logic basically) and practical approaches. detailed introduction to machine learning: "Machine Learning", Tom Mitchell. Read and understand this book, and your good to go for the next-gen a.i. system ready to rule .. err .. understand the world ;) Beware, both books are really expensive, but the best i have read about the topic. if you are more into game programming: the "A.I. Wisdom"-series (intermediate level, basically pathfinding and decision making, almost nothing about machine learning) "AI for Computer Games and Animation: A Cognitive Modeling Approach" by David Funge (a short textbook about another way of doing A.I. things in games, pressssssiuosss ;) or, if you are the "good code is better then any textbook"-type: read the podbot source (counterstrike bot) play around with "Nero - The Game" (a playground for rts-like strategical and tactical decision making and learning) and read www.gameai.com ;) thats enough for a start, i guess

[vinis]

You can learn AI just like that !!!!!! AI have many branches : Logical AI, Search, Pattern recognition, Inference, Heuristics, Genetic programming, Neural networks, Expert Systems, .... You have to buy so many books and spend so many time to learn. You choose AI when : 1.You have some problem and you want to find some smart solution. 2.You have to optimize a solution. 3.There is no other way to find a solution. The best start is to study Search algorithms and Heuristics for a problems like Turn based games. Try to implement reversi game for example and then Crossword The best Web sites are http://www.aaai.org/ http://aima.cs.berkeley.edu/ai.html The best Books for introduction to AI are : Artificial Intelligence: A Modern Approach (Russell, Norvig) Artificial Intelligence (Winston) Artificial Intelligence: Structures and Strategies for Complex Problem Solving (Luger) John Vinis The common sense is not so common...