Is there any "general" AI theory covering most paradigms?
Ondra Žižka
I am a student of IT science caught by the AI and willing for more knowledge. I hope I'll find some soulmates here.
I have learned about most types of neural networks, fuzzy systems, prolog & lisp, functional programming, genetic algorithms, evolutionary programming, etc.
But all of it seems to be just "inspired" by the nature, and is based on mathematical models. Most of them has very narrow spectrum of problems it can solve, compared to what AI is going to achieve as it's final goal (?) - universal pondering, cogitating system.
I'm not going to ask dumb questions like "how big the NN should be to learn it think" and such. My question is:
Is there some AI theory ( or idea / area of research aiming to create a theory) which would cover most currently known concepts and use them together?
What about some fuzzy graph-like database of n-tuples holding all knowledge of an intelligent system, perhaps using neural networks to create the fuzzy relations and to perform tranformations of both short-term knowledge (aka. cogitation) and long-term knowledge (learning, memorizing, creating memories) ?
Is there some such research?
I accept any references; I have a partial access to Springer, IEEE, and such.
It may be a frequent question, but in the flood of information trash all over the internet, I was not able to find something valuable.
Thanks,
Ondra Zizka
[ comp.ai is moderated ... your article may take a while to appear. ]
Dmitry A. Kazakov
> Is there some AI theory ( or idea / area of research aiming to create
> a theory) which would cover most currently known concepts and use
> them together? What about some fuzzy graph-like database of
> n-tuples holding all knowledge of an intelligent system, perhaps using
> neural networks to create the fuzzy relations and to perform
> tranformations of both short-term knowledge (aka. cogitation) and long-term
> knowledge (learning, memorizing, creating memories) ?
The topology of the graph in effect induces some distance/similarity
measure in n-dimensional space of tuples, which in turn determines how
learning works. This implies that there cannot be any universal structure,
because for any distance we could construct a problem, for which the least
distance learning will not work. Now, if the structure is to define the
distance, then that is not universal. If the distance is determined by
something else, then the structure is not *all* knowledge.
Hence there cannot be such universal thing. There could only be ones,
suitable for some class of problems. So the question is, which class of
problems is equivalent to/required by "cognition". So far, nobody knows
this.
--
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de
Andrey Gavrilov
There is not any enough appropriate universal model for simulation of mind.
But most close concept to your task is Hybrid Intelligent Systems or Hybrid
Meural Networks, in particular Fuzzy Neural Networks. Also there is concept
"Artificial General Intelligence". Search in Google. See any information on
my sites below.
Andrey Gavrilov
http://uclab.khu.ac.kr/avg
http://ermak.cs.nstu.ru
http://www.insycom.ru
Ondra Žižka
| On Wed, 07 May 2008 10:57:58 GMT, Ondra Zizka wrote:
|
| > Is there some AI theory ( or idea / area of research aiming to create
| > a theory) which would cover most currently known concepts and use
| > them together? What about some fuzzy graph-like database of
| > n-tuples holding all knowledge of an intelligent system, perhaps using
| > neural networks to create the fuzzy relations and to perform
| > tranformations of both short-term knowledge (aka. cogitation) and long-term
| > knowledge (learning, memorizing, creating memories) ?
|
| The topology of the graph in effect induces some distance/similarity
| measure in n-dimensional space of tuples, which in turn determines how
| learning works. This implies that there cannot be any universal structure,
| because for any distance we could construct a problem, for which the least
| distance learning will not work. Now, if the structure is to define the
| distance, then that is not universal. If the distance is determined by
| something else, then the structure is not *all* knowledge.
|
Sure, the structure would not hold *all* knowledge, just the storable
part of it. Having human brain as inspiration, it has specialized
centers created as described in DNA, which could hardly learn do their
task (sound processing, optical object recognition, cause/effect
principle - all these seem to be "hard-wired").
The structure would hold experience (actions done and its effects,
learned techniques), memory (remembered objects, remembered "classes
of objects", social memory), current environmental information (like
"where am I", "what's the time"), current, mid-term and long-term
"goals", etc etc.
| Hence there cannot be such universal thing. There could only be ones,
| suitable for some class of problems. So the question is, which class of
| problems is equivalent to/required by "cognition". So far, nobody knows
| this.
And any estimates?
My personal bet is that it will have something of Prolog's inference
mechanism, only the associations will be fuzzy and self-learned,
stored in the structure I've described above, and the rules will be
also subject of inference and storing - and that's the way the
intelligent system will learn:
1) current sate -> accidental actions done -> their effect ->
associations update
2) current sate -> observated environmental changes -> their effect
-> associations update
3) current sate + desired state -> inference -> assumed actions
needed -> actual effect of actions done -> associations update
Say, our intelligent system is a newborn.
Ad 1) He is hungry. Someone offers feeding. He accepts and is
not hungry. Thus, the intelligent system associates: "I am
hungry" + "Feeding" -> leads to "I am not hungry".
Ad 2) He is alone. He cries. Someone comes and offers feeding.
Associates: "I am alone" + "I cry" -> leads to "Someone comes and
offers feeding".
Ad 3) He is hungry and not alone. He wants to be fed, and by
inference, he finds the "I cry" action, with value say 0.5,
because he is not alone; it's low value, but he has nothing better,
so he tries the action "I cry". And it works! - he is fed. Thus,
updating the (fuzzy) association: "I am hungry" + "I cry" += 0.25,
"I am alone" + "I cry" -= 0.15.
Not that the newborn is powered by fuzzy prolog, but it could work
this way.
The question now is, whether such infering could solve all problems.
As far as I can imagine, it could solve quite complex tasks. What's
your opinion?
Ondra
Ondra Žižka
| "Ondra Zizka" <on...@dynawest.cz> wrote in news:48218b33$1...@news.unimelb.edu.au...
| > Is there some AI theory ( or idea / area of research aiming to create a
| > theory) which would cover most currently known concepts and use them
| > together?
| > What about some fuzzy graph-like database of n-tuples holding all
| > knowledge of an intelligent system, perhaps using neural networks to
| > create the fuzzy relations and to perform tranformations of both
| > short-term knowledge (aka. cogitation) and long-term knowledge (learning,
| > memorizing, creating memories) ?
| >
| > Is there some such research?
| > Thanks,
| > Ondra Zizka
|
| ...
| See any information on
| my sites below.
|
| Andrey Gavrilov
| http://uclab.khu.ac.kr/avg
| http://ermak.cs.nstu.ru
| http://www.insycom.ru
Thank you for the tips - based on that, I've found what I was looking
for. One of the first document's I've found is
http://www.mind-consciousness-language.com/gavrilov%20principles.pdf
There you formulate "the training principle":
"When interacting with the environment, the intelligent system
stores the associations between different images which it uses to
plan and execute its behavior..."
That is exactly what I meant that "fuzzy graph-like database of
n-tuples" would be for - it would store those associations. See my
reply to Dmitry's post.
I have some idea of a system which would combine functional
programming, evolution algorithms, neural networks, genetic algorithms
and multi-agent systems; neural networks and evolution algorithms
would be its special cases. But, it's not time for me to develop some
theories until I read what's already done in this field.
Thanks again.
Ondra Zizka
Ondra Žižka
It is very exciting to see that someone's thoughts led the same way as mine:
> Intelligent systems (natural or artificial) have a mechanism that
> selects (recognition) and activates the information resources
> (neurons, neural ensembles, frames, rules, etc.) that are essential
> to the solution of an actual task by the intelligent system, and
> that deactivates the resources that are not essential to the
> solution of an actual task.
I've hesitated to make my "crazy ideas" public, but in the light of
your work - could you have look at it? Could it work? (Note that it's
just a very rough concept without any simulations done.)
http://ondra.zizka.cz/stranky/programovani/artificial_intelligence/business_society_algorithm.texy
Thanks,
Ondra Zizka
Andrey Gavrilov
I think that your thougts cause more questions than answers.
For example, what is "enough payment to it's composing individuals"? How to
control of process of dispersing of individuals?
My opinion is following.
It is good idea to apply some concepts from AI (multi agents, genetic
algorithms and so on) to social or business systems. And this idea is not
enough novel, I think. But as I understood you try to simulate intelligent
system employing features of evaluation of social system. I think that it is
not appropriate approach, because social systems are not enough stable and
not enough smart. We have different examples from history that such system
is not learned succesfully and many positive skills are lost without
usefulness. And we can not be sure that our current trends in our
civilization is good or vice versa.
But intelligent system must be stable and have capability to learn without
missing of old knowledge.
So I recommend you to apply concepts of AI to describe and improve society
but not vice versa. Unfoirtunatley our politicians do not pay attention to
opinion of scientists.
This is just my opinion which may be not correct.
Andrey
Dmitry A. Kazakov
> "Dmitry A. Kazakov" <mai...@dmitry-kazakov.de> writes:
>| On Wed, 07 May 2008 10:57:58 GMT, Ondra Zizka wrote:
>|
>|> Is there some AI theory ( or idea / area of research aiming to create
>|> a theory) which would cover most currently known concepts and use
>|> them together? What about some fuzzy graph-like database of
>|> n-tuples holding all knowledge of an intelligent system, perhaps using
>|> neural networks to create the fuzzy relations and to perform
>|> tranformations of both short-term knowledge (aka. cogitation) and long-term
>|> knowledge (learning, memorizing, creating memories) ?
>|
>| The topology of the graph in effect induces some distance/similarity
>| measure in n-dimensional space of tuples, which in turn determines how
>| learning works. This implies that there cannot be any universal structure,
>| because for any distance we could construct a problem, for which the least
>| distance learning will not work. Now, if the structure is to define the
>| distance, then that is not universal. If the distance is determined by
>| something else, then the structure is not *all* knowledge.
>
> Sure, the structure would not hold *all* knowledge, just the storable
> part of it.
[...]
> The structure would hold experience (actions done and its effects,
> learned techniques), memory (remembered objects, remembered "classes
> of objects", social memory), current environmental information (like
> "where am I", "what's the time"), current, mid-term and long-term
> "goals", etc etc.
Well, to summarize it - this structure has no idea how to learn.
That makes your initial question meaningless. The structure without a
notion of learning is irrelevant so long it can hold all possible states of
learning . For that matter, take single integer number for a structure. It
can hold all information you have described...
> My personal bet is that it will have something of Prolog's inference
> mechanism, only the associations will be fuzzy and self-learned,
> stored in the structure I've described above, and the rules will be
> also subject of inference and storing - and that's the way the
> intelligent system will learn:
This can be disproved experimentally by constructing a problem which the
inference system cannot solve, and then presenting it to human respondents,
who would be able to solve it. (i.e. Turing test)
> 1) current sate -> accidental actions done -> their effect ->
> associations update
> 2) current sate -> observated environmental changes -> their effect
> -> associations update
> 3) current sate + desired state -> inference -> assumed actions
> needed -> actual effect of actions done -> associations update
[...]
> Not that the newborn is powered by fuzzy prolog, but it could work
> this way.
A problem with all this is in clustering the observed states (stimuli) and
actions taken into generalized/hierarchical structures of lesser
cardinality. There is a long long way between 2040x2048 pixels x 50Hz frame
rate -> "bread" -> "I am being fed."
This sort of linguistic variables construction is a part of learning, and a
subject of AI (as well as of intelligence).
> The question now is, whether such infering could solve all problems.
> As far as I can imagine, it could solve quite complex tasks. What's
> your opinion?
Certainly this cannot solve all problems. A more interesting question is
how close the class of solved problems is to "general intelligence." My
impression is that it is quite remote.
Such systems can be analysed when formalized appropriately. There exist
more and less obvious conditions for a system to work. For example, proper
identification of stimuli and actions, consistency, continuity etc.
P.S. If you get a chance to look at AI publications of 40-50s, I think you
would be surprised how close in the core were their ideas of "homeostate"
etc, to yours. It is an enjoyable reading. Unfortunately, that was and IMO
still is a wrong way. However, the idea is so attractive that I often catch
myself on thinking this way... Who knows...
--
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de
[ comp.ai is moderated ... your article may take a while to appear. ]
Ondra Žižka
| "Ondra Zizka" <on...@dynawest.cz> writes:
| > Andrey,
| >
| > It is very exciting to see that someone's thoughts led the same way as
| > mine:
| >
| >> Intelligent systems (natural or artificial) have a mechanism that
| >> selects (recognition) and activates the information resources
| >> (neurons, neural ensembles, frames, rules, etc.) that are essential
| >> to the solution of an actual task by the intelligent system, and
| >> that deactivates the resources that are not essential to the
| >> solution of an actual task.
| >
| > I've hesitated to make my "crazy ideas" public, but in the light of
| > your work - could you have look at it? Could it work? (Note that it's
| > just a very rough concept without any simulations done.)
| >
| > http://ondra.zizka.cz/stranky/programovani/artificial_intelligence/business_society_algorithm.texy
| >
| > Thanks,
| > Ondra Zizka
|
| I think that your thougts cause more questions than answers.
| For example, what is "enough payment to it's composing individuals"? How to
| control of process of dispersing of individuals?
That's right. I did not try to think out a complete system definition, only a concept bounds.
I had the same questions when writing it, but left them unanswered with a "TODO: find a mechanism".
And this TODO could be potentially solved by evolving. Although, my concept misses the description of what and how should evolve.
| My opinion is following.
| It is good idea to apply some concepts from AI (multi agents, genetic
| algorithms and so on) to social or business systems. And this idea is not
| enough novel, I think. But as I understood you try to simulate intelligent
| system employing features of evaluation of social system. I think that it is
| not appropriate approach, because social systems are not enough stable and
| not enough smart. We have different examples from history that such system
| is not learned succesfully and many positive skills are lost without
| usefulness.
I have tried to find some, but didn't find. Could you kindly give me some references, please?
| And we can not be sure that our current trends in our
| civilization is good or vice versa.
| But intelligent system must be stable and have capability to learn without
| missing of old knowledge.
| So I recommend you to apply concepts of AI to describe and improve society
| but not vice versa. Unfoirtunatley our politicians do not pay attention to
| opinion of scientists.
| This is just my opinion which may be not correct.
|
| Andrey
Not only the concept of society was an inspiration.
The "general purpose" (groups of) neurons in the brain also must specialize for some task, and then they represent the old knowledge.
Then there must be a mechanism which ensures that their specialization will not be lost, and more, that they will be found and used to solve new tasks.
This is probably done by "random" trials of using them to solve it, and depending on the result of the action, their association to a group solving the new task is strengthened [or reduced] (e.g. through the emotions concept).
The way how brain works may not be the only possible solution and we should not stick to it, but at least we know one working concept.
This may not be the way how brain works, but sound reasonably.
So, this business society model is an attempt to create a principle that could work on similar principles: allow specialization, reuse specialized units for new tasks, discard units that do not contribute to the solution of a task for other use. Maybe I should not have named it "business society".
By the way, it seems that in special case it can form a neural network and backpropagation.
I know that any concept I think out is likely to be at the level of 50's research; I am just a student after all :-) That why I asked for comments and references to the results of the research already done. Thanks for them.
Ondra
Andrey Gavrilov
> "Andrey Gavrilov" <andr_ga...@yahoo.com> writes:
> | "Ondra Zizka" <on...@dynawest.cz> writes:
I just tried to say in my comments that business or social models are not
good for usage in artificial itelligence because ones are more uncertain and
not estimateable than our mind.
Nevertheless may be you will achieve any interesting results.
Best wishes!
Andrey
Ondra Žižka
| ...
| I know that any concept I think out is likely to be at the level of 50's research; I am just a student after all :-) That why I asked for comments and references to the results of the research already done. Thanks for them.
|
| Ondra
Actually, it turns out that it's not the level of 50's but rather 80's. See Marvin Minsky: The Society of Mind (1985).