 |
|
|
 |
|
|
|
|
b2220333 |
| | | | | | posted 2/5/2012 15:26 |   | | | | | | | | | Hi,
I'm a student new from AI but in computer science background.
I'm trying to research in cognitive area by AI tech, and I prepare to take a course of AI.
I want to know what's the most important spirit concept in AI,and how to learn it correctly?
Thank you~
| | |
| | |
b2220333 |
| | | | | | posted 2/10/2012 00:56 | | | | | | | | | | Cool~~
You mention about real 3D. I think I'm just learning to use a 3D simulator gazebo with mobile robot with two hands and many useful functions like SLAM,arm Navigation,kinect,and arm picking up an object... I'm not quite sure what is 3D with mob. What cases in 3D I should prevent it first,and what is the most proper way of 3D?
Thank you~
| | |
| | |
b2220333 |
| | | | | | posted 2/10/2012 00:59 | | | | | | | | | | By the way, how these kinds of concept related with cognitive science? What I mean is how to use cognitive science to the best state on all the situation?
Thank you~
| | |
| | |
tkorrovi |
| | | | | | posted 2/10/2012 23:00 | | | | | | | | | | If you ever listen a non-peer-reviewed guy like me, i mean, you never hear it from peer.reviewed guys...
It's unrestrictedness. Think a bit rationally. Why do we need AI? To do novel and unexpected things, yes? Because conventional software is for doing expected things.
But in spite that it is evident and rational, this thought is a heresy. So be careful with it, unless you want to be a despised heretic like me, that is.
| | |  Artificial Consciousness ADS-AC project | | |
| | |
b2220333 |
| | | | | | posted 2/11/2012 01:23 | | | | | | | | | | I don't care about the thing is heretic or not. What I care is the thing is really help many many people or not. You give me a very great idea. And I continue to thinking how to make that happened, but I haven't get a actual example. Can you give me an example about 3D in AI? I will try to expand and integrate with cognitive and robotics. Thank you~
| | |
| | |
tkorrovi |
| | | | | | posted 2/11/2012 01:50 | | | | | | | | | | | |
b2220333 wrote @ 2/11/2012 1:23:00 AM:
Can you give me an example about 3D in AI?
| |
If you look at it, 3D in any computer program is not really 3D. It is amazing what illusion we see when we see 3D on a computer screen. Absolutely nothing inside the code is 3D, rather what is there is a structure, most often a tree-like structure. And this structure when rendered (traversed through node by node and any graphics formed drawn to a frame) gives us an illusion of a 3D picture. The vertices are all represented by their coordinates, which are numbers and thus a kind of structures as well. There is no 3D grid in the computer memory at all, nothing 3D in memory, only structure. This is so because any processing can be done only with such representation.
Thus it is misleading to think that when processing 3D we deal with some 3D. We deal with a structure, when moving something there, we don't touch some 3D objects in memory with an invisible finger. Rather we go through the links in the structure to find a node which touches certain coordinates.
So 3D is a kind of misleading and not the core of the problem. We have to know how to process structures, and 3D would come from there. The same as it is in human brain, likely. So things more fundamental than 3D have to be dealt with first, and 3D is not the key for solution of the fundamental problems.
But consider that i deal only with True AI. So i described it fundamentally, if it is different in some conventional AI program, others should tell that.
| | | Artificial Consciousness ADS-AC project | | | Last edited by tkorrovi @ 2/11/2012 1:58:00 AM | | |
| | |
tkorrovi |
| | | | | | posted 2/11/2012 04:10 | | | | | | | | | | | |
jeremy duncan wrote @ 2/11/2012 2:10:00 AM:
And in case you don't know, if you read the tkorovvi belief, a program should learn by germination process but over many hundreds of years to get to what I describe as job 3.
| |
No not exactly many hundreds of years, the task is to research how it develops. And there is no magic, why it should develop is rationally explained, and unrestrictedness means that whatever system can emerge during the self-development. This and the theory around it supposed to be, and in many discussions here has proved to be, the center point of the research, and in a way also AI research. This system was derived from theoretical requirements, and at least there are not many possibilities for deriving such system.
| | | Artificial Consciousness ADS-AC project | | | Last edited by tkorrovi @ 2/11/2012 4:12:00 AM | | |
| | |
b2220333 |
| | | | | | posted 2/11/2012 05:07 | | | | | | | | | | Thank you(everyone) for attention.
Let me reply one by one.
First is to tkorrovi.
You describe 3D very clearly. I think you want to do the actual 3D actions not just pretend to or looks like.
Now the idea come to my mind is I can use 3D simulator like gazebo or openrave and implement AI techs on those simulator. So I want to ask more and more detailed question like:
1. What area that illusion AI couldn't reach it and should use actual 3D AI?
2. Is 3D simulator(gazebo,openrave,...) suitable for your concept? Or should I restart a totally new 3D simulator for AI?
3. What kinds of libraries we should depend on for building actual 3D AI?
Thank you~
| | |
| | |
b2220333 |
| | | | | | posted 2/11/2012 05:15 | | | | | | | | | | PS.
This is the youtube video for linking actual robot and simulator robot using PR2 robot and Gazebo simulator.
http://www.youtube.com/watch?v=qziUJcUDfBc
| | |
| | |
b2220333 |
| | | | | | posted 2/11/2012 05:41 | | | | | | | | | | Now for jeremy duncan:
Sorry for honest,I am not quite understand your ideas.
I have another maybe anyone who interest in actual 3D idea can join this chat ( I have created that room):
http://chat.stackexchange.com/rooms/2457/actual-3d-on-ai-and-cognitive-science
And I also reply detailed below:
3D
Job 1.) A observes B, B is a command for A.
In 3D this is distance of the eyes to the screen surface.
>> I think this is built by 2 man model and stand on different location and not too far from each other.
Job 2.) Job 1 goes to job 2, does job 2 keep the command?
Job two is to be ordered and keep that order.
is the 3D fuller than it should be? Are the faces giants faces?
>> 1. What means that 'does job 2 keep the command' ?
2. Is job2 means that two person come nearly?
3. How 3D full?
Job 3.) Job 2 goes to job 3, does job 3 keep the command?
If job 1, 2, 3 are all kept the task is complete.
The 3D moves, is it a jumpy 3D camera or a smooth moving 3D camera?
because the movement can break the 3D operation if the camera is all jumpy.
>> I can understand job 3,but I don't understand why you want to provide these 3 jobs. What's these for?
"how these kinds of concept related with cognitive science?"
You create a point, and the other point moves, as it moves it emulates the first point.
This is the catalyst, the prime force behind the concept, like germination.
Then you build a simple function.
Then you build a complex function.
It's the principle of evolution, you germinate, or leaven the dough, then start small then build up from there to get where you want to be.
>> This idea(germination) is very important concept, I will keep in mind. And one small question is what methods and how many methods you will choose maybe suitable for implement this kind of concept?
A mob from a team: a team operates a complex function, but as it gets from job 1 to job 3 it has failed and loses it's motivation to act professional.
A child is trained and as he or she is introduced to other children's behavior they learn to cheat, so job 3 is not the quality their parents ingrained into them.
It looks nice in appearance, but inside it's distorted and it moves or behaves erratically.
>> I know this idea you want to say is they not actually done for good jobs and act professional. I'm not quite understand that why you use 'cheat' word. Is someone cheat in real world?
tkorrovi, please edit your wiki to make it read less like a patent and more like a human readable description.
Edit,
And in case you don't know, if you read the tkorovvi belief, a program should learn by germination process but over many hundreds of years to get to what I describe as job 3. This is his idea of true AI, germination that somehow gets to what I describe as job 3 in some magical way he describes as unrestricted. But I think I describe here his idea too, but maybe he will describe how this description isn't his SW idea too?
>> Thank you for providing more ideas for talking more details.
| | |
| | |
tkorrovi |
| | | | | | posted 2/11/2012 08:16 | | | | | | | | | | | |
b2220333 wrote @ 2/11/2012 5:07:00 AM:
I think you want to do the actual 3D actions not just pretend to or looks like.
| |
Actual.
It seems we are talking about different things here. gazebo and openrave are simulators for testing robots in a computer, so that they "live in" an artificial 3D environment. Such environments can be created with a 3D library, as we can put camera everywhere (where we project the things) and feed the image to the robot as a 2D image.
But these are completely different things. I talked about an actual modeling of the environment in 3D by an artificial system.
About 3D libraries, i don't know them all and different libraries are necessary for different purposes. I think open source libraries should be preferred, also with these it is possible to use some code from the library to create new code. I also cannot give any one answer what 3D library should be the best for AI. I have used OpenSceneGraph which is a good general purpose 3D library. It is mostly good though for 3D models which consist of solid 3D objects, for other objects and effects it is not that good.
| | | Artificial Consciousness ADS-AC project | | | Last edited by tkorrovi @ 2/11/2012 8:20:00 AM | | |
| | |
b2220333 |
| | | | | | posted 2/11/2012 09:14 | | | | | | | | | | You talked about an actual modeling of the environment in 3D by an artificial system. I have two questions:
1. What area that illusion AI couldn't reach it and should use actual 3D AI?
2. In your idea, seems if this idea become true,it will lead a revolution in AI, how it possible?
After really know the purpose of Actual AI and how to make it happened, I will begin to learn any methods useful to figure out what we can done for it.
Thank you~
| | |
| | |
b2220333 |
| | | | | | posted 2/11/2012 10:43 | | | | | | | | | | Sorry for my poor english.
I try my best to understand what you ask.
I think you're right,if I just answer your question, then you can have a way to express what you thinking.
You don't have to care about the tone when you are genuine.
I will always listen to year ideas if you willing to say, and try my best to understand the words you told to me.
----Here is the start---
1.) Why is the distance of the eyes to the screen important?
Is it because of two robots standing next to each other?
If no, then why?
=>Because of you said: Job 1.) A observes B, B is a command for A.
In 3D this is distance of the eyes to the screen surface.
In my opinion,A stand for a human model or robot,and B also too. Then I know real 3D,but I don't know which distance do you say about. I just guess maybe in a actual 3D simulator, and the distance is created by A,B's eyes to the simulator's camera. So distances are two lines.
I know the distance which between camera and A,B is important,but I can't understand that distance is what you think or not. You don't want me to ask, so I can stop asking to you.
2.) Why does 3D project faces to be giants faces, and people to be extremely small people? The problem is the way they were filmed, but what is the problem and how do you fix it?
=> Oh, I suddenly aware maybe I understand what you mean in this issue. In simulated 3D,some not important part is just ignore or some important part are specially be enlarged. If using actual 3D, it will not add any man-made effect but also honestly display the model itself.
3.) What problem does a fast jumpy camera present to stereoscopic 3D that ruins the 3D experience?
=>To be honest, in my knowledge, I have no ability to answer it. I just guess maybe it's a sampling problem.
Sorry to let you disappointed.
Remember I said in my first post to you, that You need to know 3D and you say what so I show you what you need to know, then you go and say "do you mean two robots standing beside each other?". When i read you say that about the 3D I was explaining I just about ignored you, but I came back to re-explain myself.
=>Thank for your patient, if you will to talk more,I will learn from you in my whole life; if you wont, that's ok, just relax~ I do not want to make anyone annoyed.
These three things present the base to complex pattern that I describe as AI.
You asked for the spirit or principle of AI and I thought that was all you were asking for, so I told you.
=> Thank you for providing ideas, even now I can't understand that totally or just thinking a mess,I'm also glad to have this conversation.
If you can present your AI in the three steps I mentioned, from base to complex, then you will be accomplished if it works.
=> Thanking for this hint.
Now answer my questions before I answer any more of yours. And don't say do you mean this or that, Google the answer and it should be something to do with 3D and not two robots standing beside each other.
Sorry for the tone, but that threw me for a loop when I read that.
=> If you also feel any unconfortable, just feel free to tell me know,I will try to self-criticism myself.
| | |
| | |
tkorrovi |
| | | | | | posted 2/11/2012 18:14 | | | | | | | | | | | |
b2220333 wrote @ 2/11/2012 9:14:00 AM:
| |
> 1. What area that illusion AI couldn't reach it and should use actual 3D AI?
I did not talk about the actual AI, i just explained what the 3D is all about and this also should say something about how to model it in AI. I only deal with True AI as i said. And i described how the actual 3D works, the 3D libraries may or may not be used for modeling the environment in AI, just the principles are the same.
> 2. In your idea, seems if this idea become true,it will lead a revolution in AI, how it possible?
The first thing i asked you was whether you ever listen a non-peer-reviewed guy like me. You already fell to the hole and listened to me. Now you start to think whether this can be true, whether it is possible that a non-peer-reviewed guy can know something important. I asked you this first, so you had to think about it first, before listening to anything what i say.
What concerns heresy then i said that also in purpose. In time you would understand it.
Yes it will lead to revolution, and not only in AI. Greater than anything man has ever done. What concerns AI, all these principles seem to be valid for all AI, also for conventional AI. In that all AI systems are in a way variations of the same thing, though in some restricted way. In a way that there is a common theory valid for all AI, but this is only about the most fundamental, most of the particulars of differentkind of AI systems cannot be derived from this. Because they are very specific for the task for which such AI systems are made for.
It is not that i alone have thought out something great, and then this is not probable and thus must be completely dismissed. It is not me alone, i only developed that which some people have thought out already a long time ago. The only problem is that you have not heard about it, because it is not official, because it is heresy today. All science started from heresy. Giordano Bruno, Galileo Galilei, Isaac Newton, all were hermeticists. And now when we want to go further, we must go back to where everything started.
You may think i go too far. But what it is about is True AI. Think what it is, the most profound thing there is, so beyond any doubt it should be about the most fundamental. And thus everything we know about the most fundamental should apply. So no wonder it is related to hermeticism, philosophy, all what humans have ever found out about the most fundamental things. Who say no, it is not so, True AI is something much easier, it is a simple to make it or find it somewhere without so much thinking, or it is just about doing and not thinking, and one does not have to think about so deep things to understand it, would you believe them?
Also what concerns True AI, one cannot make a "simple version" of it which is still True AI. One can make True AI to do only simple things, but not a simple version of a True AI. Because what it is all about is unrestrictedness. Whenever you simplify the core of it, you restrict it, and even a small restrictions cause severe inability to cope with huge amount of things, resulting in exceptions and inconsistencies. I call such simplifications destructive simplifications, one example is artificial neural networks, when they are intended to be used as a True AI. What concerns True AI all such things are partly right, but partly not, and partly restricted.
I'm sorry for talking so much about True AI. But this is the only thing i deal with. You may say you want to make a conventional AI and thus it is unnecessary to talk about all that here. But i'm telling all that just because you may find some of it useful for conventional AI. Because after all it is about the most fundamental in AI.
| | | Artificial Consciousness ADS-AC project | | | Last edited by tkorrovi @ 2/11/2012 6:18:00 PM | | |
| | |
b2220333 |
| | | | | | posted 2/11/2012 19:11 | | | | | | | | | | Let me share my feeling~
You don't have to say sorry. I appreciate your ideas. You want to share to a stranger also prove you are so nice.
This sentence from you I like it so much : One can make True AI to do only simple things, but not a simple version of a True AI.
I also believe that some people can understand something without trying to thinking, because I have a part of that skills. I feel that true AI is important because it's not derived by some formula like thinking but using a very simple mechanism and a very very large scale particle compose together to form a system which can learn,thinking,adjust itself,...etc.
If that 3D possible,I think,maybe needs a large amount of GPUs and the most important thing is to have the ability to know how the basic particle works. Is that right?
Any suggestion also be welcomed. Please feel free to say anything~
Thank you~
| | |
| | |
tkorrovi |
| | | | | | posted 2/12/2012 19:21 | | | | | | | | | | Jeremy, your post does not belong here, would you please move it to your own thread. Please try to be more concise here, people cannot find your answers to questions from your long text.
Thought about creating a site somewhere called AI Heretics, in the form of wiki, about AI based on hermeticism. That is not real now though, not enough people interested for it to become popular. So just remember the idea.
| | | Artificial Consciousness ADS-AC project | | |
| | |
b2220333 |
| | | | | | posted 2/13/2012 17:22 | | | | | | | | | | @jeremy duncan
Thank you for very very detail explanation.
I am shocked when I see it first time, it seems an essay.
You are so great to create so long article in short time.
And though many many examples and final picture, I suddenly realize maybe I know what your 3D means, even I can't understand those example clearly and compose together(maybe to far from my concept). I think you maybe want to create a special 3D simulator let anyone who see that 3D scene will move when the observer's eye move. It compose a real scene just like in our life. Is that right? Any comment will be appreciated. Thank you~
| | |
| | |
b2220333 |
| | | | | | posted 2/13/2012 17:31 | | | | | | | | | | @tkorrovi:
Thank you for understanding my poor knowledge of AI tech.But I'm really want to learn and do the most powerful thing to really help anyone in the world. Any further ideas I will appreciated, and try to figure a way which I can go for it. Now, I hear two different version of 3D. I want to learn both because in some cases they can combine together. In your idea, I want to know how to really implement true AI using 3D. And how could I go further?
| | |
| | |
tkorrovi |
| | | | | | posted 2/13/2012 21:59 | | | | | | | | | | | |
b2220333 wrote @ 2/13/2012 5:31:00 PM:
I want to know how to really implement true AI using 3D.
| |
I don't think True AI can be implemented using 3D. I think True AI can be implemented, and then trained to perceive 3D. Because the structures there are are also structures which are necessary for modeling 3D. But there cannot be 3D in the basic mechanism, because 3D or any fixed number of dimensions is not fundamental, only a structure (system) is fundamental, and thus any 3D on the basic level restricts. There cannot be anything non-fundamental in True AI.
| | | Artificial Consciousness ADS-AC project | | |
| | |
b2220333 |
| | | | | | posted 2/14/2012 04:02 | | | | | | | | | | Now some ideas came to my mind.
I try to describe the following things. If it has any error or misunderstanding, just feel free to say more and more.
True AI: The simulator can really simulate all human brain activities.
(Maybe can simulated by cognitive architecture)
3D (four different way):
1. Outside world we live in.
2. simulated like 3D (this is like gazebo)
3. The 3D scene in simulator can be changed by moving of observer.
4. Actual 3D simulator (means all stuff are compose by particles and waves to really close to the world's physics we live)
If these definition are no problems, which one you think is the most important to implement first? And which one is deserve to devote in whole life?
Thank you~
| | |
|
|
|
|
|
