Re: [computer-go] Slides for Villach-EC Lecture
The word in AI I don't feel conformable with is Artificial, not Intelligence. I use _Abstract_ Intelligence (also AI) as a replacement. Have you ever heard of artificial aerodynamics (applicable to planes but not to birds) or artificial thermodynamics (the same). I understand that AI is the science of thinking machines and that, of course, applies to biological ones. Aerodynamics was not developed by ornithologists and neither will the theory of thinking machines be developed by any frog-rippers. Electrical engineers laugh at a joke in which someone pretends to have found The Physical Location Of The Notepad (all capitalization is insufficient to reach the pompous statements frog-rippers use to make) in the floppy disk. After a conclusive experience placing electrodes on different parts of a computer, it has been proved that the floppy disk unit triggers each time the application Notepad is launched. Of course, the reason behind the floppy access is the last file opened was a:\readme.txt. All serious science is a part of Mathematics ( if not, Mathematics will absorb it ;-) ) and so is the science of thinking machines: AI. Jacques. PD. The only explanation I find to the need of the word artificial is the (for me totally unexplainable) respect educated Anglo-Saxons have for religion. This way, it looks compatible with religion, but of course it isn't. The soul of a thinking machine is an anti-scientific notion no matter what kind of thinking machine. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Slides for Villach-EC Lecture
On Sun, 2007-07-22 at 12:59 +0100, Jacques Basaldúa wrote: PD. The only explanation I find to the need of the word artificial is the (for me totally unexplainable) respect educated Anglo-Saxons have for religion. This way, it looks compatible with religion, but of course it isn't. The soul of a thinking machine is an anti-scientific notion no matter what kind of thinking machine. I think you are looking to fit a very complicated explanation to this when a very simple one makes the most sense.Artificial is the term we always use in the English language to denote something man made that happens to have a counterpart in nature. I don't understand the need (to use your phrase) to put the blame on religion in this case. However I DO understand why you might have an axe to grind with religion, it has been the cause of so much misery and world problems - but that's off topic if anything ever was! Just because the word artificial was used doesn't mean there was an ulterior motive or need to justify something.Don't forget this word was coined many decades ago, usually with experience and familiarity with something we find better terminology that is perhaps more appropriate (or more politically correct) but in my opinion the term was very well chosen. - Don ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Slides for Villach-EC Lecture
Hi Chrilly, I am sorry about your fight with a dog, and I hope you are ok! I read your slides: interesting point of view, whereas you seem a little frustrated. Thank you for sharing your opinion. However, I have to disagree with this statement: UCT: Complete Antithesis to AI-approach I really thing it is exactly a modern AI approach!! Also it is a general algorithm applied to many different domains (and many are not two player games, ie max-max problems and not min-max). I think it is exactly the bad example for the anti-drosophila thesis... Cheers, Sylvain 2007/7/21, chrilly [EMAIL PROTECTED]: Attached are Powerpoint-Slides for a computer-Go-lecture I should have given tomorrow (Sunday 21.07.07) at the European-Championship in Villach/Austria. I think the final conclusion fits to the cancellation of the Gifu tournament. I did not know this when preparing the slides. Unfortunately I had to cancel the lectures, because at an attempt to stop the fighting between my dog Bello and his village-rival Max I was considerable bitten by Max. Normally the owner of Max and myself let them fight. It looks like they would kill each other, but in fact nothing serious happens. But this time I wanted to play the hero and Max was not pleased that I interferre into dog-internal affairs (neither was Bello). Maybe the slides are of interest. Chrilly ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Slides for Villach-EC Lecture
However, I have to disagree with this statement: UCT: Complete Antithesis to AI-approach Martin Mueller quotes J.McCarthy in his thesis: The research of Go programs is still in its infancy, but we shall see that to bring Go programs to a level comparable with current Chess programs, investigations of a totally different kind than used in computer chess are needed. UCT is different to Alpha-Beta (not totally, because its some other form of search, but it is different). I am sure, McCarthy had not UCT in mind. It was always the goal of McCarthy and his followers to simulate and to surpass the human mind. HAL in Stanly Kubrics Odyssee in space 2001 is the dream-computer of this discipline. UCT has nothing to do with human Go. It has some similarity to the behaviour of ant-collonies (its not in the technical sense an ant-colony algo). It was never the goal of AI to explain ants. I really thing it is exactly a modern AI approach!! Also it is a general algorithm applied to many different domains (and many are not two player games, ie max-max problems and not min-max). I full aggree, it is a general and very interesting algorithm which can be applied to many domains. How would you define modern AI? Obviously it is not the classic approach to mimic humans anymore. But what is it? In my opinion is UCT a statistical estimation method. The armed-bandit is classical statistical problem. I think it is exactly the bad example for the anti-drosophila thesis... What do we learn about the human mind from UCT? Chrilly ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Slides for Villach-EC Lecture
Hi Chrilly, It was always the goal of McCarthy and his followers to simulate and to surpass the human mind. (...) UCT has nothing to do with human Go. It has some similarity to the behaviour of ant-collonies (its not in the technical sense an ant-colony algo). It was never the goal of AI to explain ants. Ok I understand your point. For me the goal of AI is not to explain the human mind, and even less to try to imitate it. Intelligence is not about how it works, but what it does. If it seems intelligent, then it is. Are animals and human intelligent? They seem so they are: it is simply an ill-defined question. How would you define modern AI? Obviously it is not the classic approach to mimic humans anymore. But what is it? For me it is when we (I was not there :-)) become less philosophical and more precise about what we want. We want a system which use data to improve itself in order to adapt to unseen situations. What is a good learning? In supervised learning we have some answers, as what is over-fitting, how to avoid it, how to handle well non linear and structured data (e.g. with kernel methods). In control problem, ie reinforcement learning, which is for me the big challenge, and were big applications are, many questions are still fuzzy, but modern AI is for me trying to make them well-posed, not doing philosophy on how intelligence could arise, or how it works in human mind (which is so complicated). What do we learn about the human mind from UCT? Nothing and that's not the goal, I simply don't care. If you really want to learn from human mind, do cognitive science, not AI. Maybe one interesting thing we can learn is that there is not only one way to make things intelligent, but many. Sorry to have been somehow philosophical here, I'll stop :-). Sylvain ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Slides for Villach-EC Lecture
I generally also disagree with your statement - but I would point out that it's difficult to reason about this without a clear definition of AI. AI gets redefined every time the magic is understood. I think it largely comes down to semantics. UCT fits just about any definition of AI more than straight alpha beta mini-max search does - but of course alpha beta was once considered mainstream AI and now has little respect because it's no longer cutting edge. UCT fits the human brain model much more closely than standard mini-max search. A few days ago I posted something about this. UCT simulates intelligence better than alpha beta with evaluation function because it has the ability to learn on the fly without explicitly coded knowledge. And UCT programs play as if they actually understand the game better - even if they are a little weaker tactically. The approach of building an evaluation function with a zillion of if then rules and tables to me is not AI. Ok, it really IS AI in the classic sense but if anything needs to be redefined this does! The UCT search itself is closer to the human way too in my opinion. Although alpha beta can still be highly selective, the best first style of UCT is closer to what we do. Moves that do not appear to make sense get very little consideration. If you look under the hood, you can argue that there is no such thing as AI, after understanding how a piece of software works you see that ultimately it's just a program, a fixed set of instructions and nothing more than a hard coded set of if then rules. But what we call AI is anything that we can code that is sufficiently complex that we can imagine (if we cross our eyes) it is actually smart. 60 years ago multiplying 2 large numbers together with a machine was considered a highly intelligent task which proved intelligence. - Don On Sat, 2007-07-21 at 14:40 +0200, chrilly wrote: However, I have to disagree with this statement: UCT: Complete Antithesis to AI-approach Martin Mueller quotes J.McCarthy in his thesis: The research of Go programs is still in its infancy, but we shall see that to bring Go programs to a level comparable with current Chess programs, investigations of a totally different kind than used in computer chess are needed. UCT is different to Alpha-Beta (not totally, because its some other form of search, but it is different). I am sure, McCarthy had not UCT in mind. It was always the goal of McCarthy and his followers to simulate and to surpass the human mind. HAL in Stanly Kubrics Odyssee in space 2001 is the dream-computer of this discipline. UCT has nothing to do with human Go. It has some similarity to the behaviour of ant-collonies (its not in the technical sense an ant-colony algo). It was never the goal of AI to explain ants. I really thing it is exactly a modern AI approach!! Also it is a general algorithm applied to many different domains (and many are not two player games, ie max-max problems and not min-max). I full aggree, it is a general and very interesting algorithm which can be applied to many domains. How would you define modern AI? Obviously it is not the classic approach to mimic humans anymore. But what is it? In my opinion is UCT a statistical estimation method. The armed-bandit is classical statistical problem. I think it is exactly the bad example for the anti-drosophila thesis... What do we learn about the human mind from UCT? Chrilly ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Slides for Villach-EC Lecture
Chrilly, your definition of AI is too limited. See, for example, http://www.cse.buffalo.edu/~rapaport/definitions.of.ai.html. Regards, Hideki (gg) chrilly: [EMAIL PROTECTED]: However, I have to disagree with this statement: UCT: Complete Antithesis to AI-approach Martin Mueller quotes J.McCarthy in his thesis: The research of Go programs is still in its infancy, but we shall see that to bring Go programs to a level comparable with current Chess programs, investigations of a totally different kind than used in computer chess are needed. UCT is different to Alpha-Beta (not totally, because its some other form of search, but it is different). I am sure, McCarthy had not UCT in mind. It was always the goal of McCarthy and his followers to simulate and to surpass the human mind. HAL in Stanly Kubrics Odyssee in space 2001 is the dream-computer of this discipline. UCT has nothing to do with human Go. It has some similarity to the behaviour of ant-collonies (its not in the technical sense an ant-colony algo). It was never the goal of AI to explain ants. I really thing it is exactly a modern AI approach!! Also it is a general algorithm applied to many different domains (and many are not two player games, ie max-max problems and not min-max). I full aggree, it is a general and very interesting algorithm which can be applied to many domains. How would you define modern AI? Obviously it is not the classic approach to mimic humans anymore. But what is it? In my opinion is UCT a statistical estimation method. The armed-bandit is classical statistical problem. I think it is exactly the bad example for the anti-drosophila thesis... What do we learn about the human mind from UCT? Chrilly inline file ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ -- [EMAIL PROTECTED] (Kato) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Slides for Villach-EC Lecture
How would you define modern AI? Obviously it is not the classic approach to mimic humans anymore. But what is it? For me it is when we (I was not there :-)) become less philosophical and more precise about what we want. We want a system which use data to improve itself in order to adapt to unseen situations. In this sense the Elo-System is an AI algorithm. If you feed more data/games the quality of prediction increases. It is in fact a weakness of the Elo-Rating that this is not taken into account (newer systems like TrueScore do). Remi used an Elo-Rating (Bradley-Terry model) for his pattern classifier. In this broad sense UCT is AI, but I would classify it as a branch of applied statistics. Bradley-Terry was invented long before the name AI was coined. What do we learn about the human mind from UCT? Nothing and that's not the goal, I simply don't care. You are already fallen from grace :-) You just want to make a strong Go-programm and develop some good - general purpose - algorithms :-). (There is a famous article of DonskeyJ.Schaeffer with the title Falling from Grace in 1989). Chrilly P.S.: I am frustrated in the sense that you book for holiday a hotel and when you arrive, you see that it is - according your standard - only **. This has in my case a very positive and not at all frustrating background. I get in the moment a lot of interesting and well paid offers for contracts (its a little bit selling the Hydra-fame). Humans adapt quick to higher standards and I am therefore not satiesfied anymore with the Go-hotel I have booked last year. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Slides for Villach-EC Lecture
Below is my favorite one in the list. An example of this are neural-networks. Neural networks are just a parameter-free optimization/estimation method. No magic at all, just a boring and not very efficient estimator. Chrilly Warning: Cynical Definition... My definition of AI is any algorithm that is new in computer science. Once the algorithm becomes accepted then it's not AI, it's just a boring algorithm. At one time windows, mouse, menus, scroolbars etc. were considered an AI technique for makeing computers understand natural language. (The menus are a list of valid words the system understands) This is also why I study Cognition, not AI. R. Keene Chrilly, your definition of AI is too limited. See, for example, http://www.cse.buffalo.edu/~rapaport/definitions.of.ai.html. Regards, Hideki (gg) chrilly: [EMAIL PROTECTED]: However, I have to disagree with this statement: UCT: Complete Antithesis to AI-approach Martin Mueller quotes J.McCarthy in his thesis: The research of Go programs is still in its infancy, but we shall see that to bring Go programs to a level comparable with current Chess programs, investigations of a totally different kind than used in computer chess are needed. UCT is different to Alpha-Beta (not totally, because its some other form of search, but it is different). I am sure, McCarthy had not UCT in mind. It was always the goal of McCarthy and his followers to simulate and to surpass the human mind. HAL in Stanly Kubrics Odyssee in space 2001 is the dream-computer of this discipline. UCT has nothing to do with human Go. It has some similarity to the behaviour of ant-collonies (its not in the technical sense an ant-colony algo). It was never the goal of AI to explain ants. I really thing it is exactly a modern AI approach!! Also it is a general algorithm applied to many different domains (and many are not two player games, ie max-max problems and not min-max). I full aggree, it is a general and very interesting algorithm which can be applied to many domains. How would you define modern AI? Obviously it is not the classic approach to mimic humans anymore. But what is it? In my opinion is UCT a statistical estimation method. The armed-bandit is classical statistical problem. I think it is exactly the bad example for the anti-drosophila thesis... What do we learn about the human mind from UCT? Chrilly inline file ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ -- [EMAIL PROTECTED] (Kato) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Slides for Villach-EC Lecture
On Sat, 2007-07-21 at 17:37 +0200, chrilly wrote: In this sense the Elo-System is an AI algorithm. I found this interesting site that may be vaguely related - it implies that ELO is a better way to measure intelligence and provides a methodology: http://home.earthlink.net/~bmcgaugh/eloiq.htm - Don ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Slides for Villach-EC Lecture
chrilly: [EMAIL PROTECTED]: Below is my favorite one in the list. An example of this are neural-networks. Neural networks are just a parameter-free optimization/estimation method. No magic at all, just a boring and not very efficient estimator. It's not neural networks but just a perceptron with backpropagation learning algorithm. There are lots of neural network models in the world. Some are developed by neuroscientists or cognitive scientists. #Even backpropagation is used in some real applications, Dr. Amari wrote. Chrilly Warning: Cynical Definition... My definition of AI is any algorithm that is new in computer science. Once the algorithm becomes accepted then it's not AI, it's just a boring algorithm. I agree half of this. Not all algorithms derived from AI are boring :-). At one time windows, mouse, menus, scroolbars etc. were considered an AI technique for makeing computers understand natural language. (The menus are a list of valid words the system understands) Really? I don't know such bizarre ad. This is also why I study Cognition, not AI. R. Keene Regards, Hideki Chrilly, your definition of AI is too limited. See, for example, http://www.cse.buffalo.edu/~rapaport/definitions.of.ai.html. Regards, Hideki (gg) chrilly: [EMAIL PROTECTED]: However, I have to disagree with this statement: UCT: Complete Antithesis to AI-approach Martin Mueller quotes J.McCarthy in his thesis: The research of Go programs is still in its infancy, but we shall see that to bring Go programs to a level comparable with current Chess programs, investigations of a totally different kind than used in computer chess are needed. UCT is different to Alpha-Beta (not totally, because its some other form of search, but it is different). I am sure, McCarthy had not UCT in mind. It was always the goal of McCarthy and his followers to simulate and to surpass the human mind. HAL in Stanly Kubrics Odyssee in space 2001 is the dream-computer of this discipline. UCT has nothing to do with human Go. It has some similarity to the behaviour of ant-collonies (its not in the technical sense an ant-colony algo). It was never the goal of AI to explain ants. I really thing it is exactly a modern AI approach!! Also it is a general algorithm applied to many different domains (and many are not two player games, ie max-max problems and not min-max). I full aggree, it is a general and very interesting algorithm which can be applied to many domains. How would you define modern AI? Obviously it is not the classic approach to mimic humans anymore. But what is it? In my opinion is UCT a statistical estimation method. The armed-bandit is classical statistical problem. I think it is exactly the bad example for the anti-drosophila thesis... What do we learn about the human mind from UCT? Chrilly inline file ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ -- [EMAIL PROTECTED] (Kato) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ -- [EMAIL PROTECTED] (Kato) ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Slides for Villach-EC Lecture
On Sun, 2007-07-22 at 02:02 +0900, Hideki Kato wrote: Warning: Cynical Definition... My definition of AI is any algorithm that is new in computer science. Once the algorithm becomes accepted then it's not AI, it's just a boring algorithm. I agree half of this. Not all algorithms derived from AI are boring :-). Boring is a subject terms - but this quote hits the nail on the head. Anything can be boring to you once you understand it so well the magic is gone. I used to think alpha beta pruning was a magical wonder. - Don ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Slides for Villach-EC Lecture
On 7/21/07, chrilly [EMAIL PROTECTED] wrote: If you feed more data/games the quality of prediction increases. It is in fact a weakness of the Elo-Rating that this is not taken into account (newer systems like TrueScore do). Can you provide a link to TrueScore? My searches are coming up empty. - Brian ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Slides for Villach-EC Lecture
Sorry, it is TrueSkill and not TrueScore. http://research.microsoft.com/mlp/trueskill/ Chrilly - Original Message - From: Brian Slesinsky [EMAIL PROTECTED] To: computer-go computer-go@computer-go.org Sent: Saturday, July 21, 2007 9:10 PM Subject: Re: [computer-go] Slides for Villach-EC Lecture On 7/21/07, chrilly [EMAIL PROTECTED] wrote: If you feed more data/games the quality of prediction increases. It is in fact a weakness of the Elo-Rating that this is not taken into account (newer systems like TrueScore do). Can you provide a link to TrueScore? My searches are coming up empty. - Brian ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Slides for Villach-EC Lecture
Warning: Cynical Definition... My definition of AI is any algorithm that is new in computer science. Too much of an assumption here: That we should be designing using _algorithms_! A good piece of what puts a problem/program in the realm of AI is the need to use heuristics. The problems we really want to work on are less likely to have foolproof solutions, more likely to require fudgework. Go is way past the borders of Solution Country. We can't strictly mimic human ways of choosing moves, but we have to recognize that nothing else we have comes close. The explosively growing search space in any midgame position, the sensitive dependence of outcomes on move-order, the bewildering number of possible configurations of stones--in which transposing, changing one stone's color, or even moving it one space (or even putting the exact same pattern into a different arrangement of neighboring groups) can entirely change the situation I think we need to design and breed creatures that live in Goworld. In other words, write programs that turn simple codes into structures they can interpret as methods for generating moves--then test these creatures against one another, exchange elements of the better ones, go on choosing more successful configurations. We'd start with simple, clumsy structures barely adequate to play each other. If the elements are well-designed to link together in plausible ways, though, we ought to end up with structures able to do things we'd find it hard to specify--and to choose which of them are appropriate in any particular position, more easily than we could specify appropriate conditions for them. Could such things embody higher-order concepts--like live group? I think they'd prefer moves (good shape) that readily formed live groups, or linked precarious groups to live ones, or interfered with an opponent's efforts to do so. Would the program need to run four billion years? Or a shorter, equally impractical time? I really don't know, but I'm inclined to think it worth a try. Easier than telling a machine, say, how to know whether a group is live, when it needs to be defended, when it should be abandoned--or sacrificed--or when half of it might get more territory by forsaking the rest and linking up to other neighbors. Even good human players are not foolproof; they make patterns that don't use many stones, are likely to be easily defensible, are likely to be good bases for expansion harrassment of the opponent. They First build a little house, like the Korean Baptist minister used to tell me... Forrest Curo - This email was sent using AIS WebMail. http://www.americanis.net/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Slides for Villach-EC Lecture
Can computer intelligence replace that of human? ? This question comes up often. Actually it's not a philosophical question,because it has an exact scientific answer. Even though this answer is based on an important scientifc axiom that I will mention. So a computer can lock human out of the control. It's no big?deal, my door knob locked me out of my house all times. It can reason too and adapts the best policy of anti-persuation: silence. Let's take an example to explain the question.?The 128-bit password. In theory no existing and forseeable computer can solve the password. However, on average many poeple?use names or dates as the password. Incorporating these data a computer has a large?probablity to solve some passwaords, just as a human can guess it out, and do it better than human.??However, a human has to program these data into the computer. But, we can write a program that can find this connection all by itself.?To write such a program one?cannot?set out to write a decyphering program, instead one has?to write a program to describe everything in the life of George, the guy whose password we want to decypher. Password decyphering is only part of the calculations the program comes up by all itself. The moral of this example is that to solve one problem all by the program itself, the program has to be more general, or covers a larger and more fundamental domain, than the problem itself.?Does a domain exis ts that covers everything in this universe and others? Someone may point out immediately?that a computer?program does not have to cover everything in the universe. All it needs is to cover more than the domain?as human do to exceed the human intelligence.??Ok. But this is not the whole story. A computer program by itself cannot develop such intelligence, because it needs data and interaction with the rest of world. The question becomes could a computer-robot combination?replace human intelligence? With the computer thinking and the robot do whatever it wants. This sounds a winning combination.?For example, the computer can instruct the robot to build the power plant,mining resources, manufacture more powerful computers,and developmore powerful programs. Why can't this happen? It's here comes?an axiom. That is an intellectual perpetual machine cannot be built.?? DL AOL now offers free email to everyone. Find out more about what's free from AOL at AOL.com. ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
