From: andr_gavriloov@yahoo.com
"Dmitry A. Kazakov" writes:
> On Sun, 25 May 2008 00:52:35 GMT, Andrey Gavrilov wrote:
>
>> I would like not say that it is fundamenrtal law of thinking. I would
>> like
>> say that it is property of thinking.
>
> OK, is it a universal property of?
>
>> And why you are sure that quantum physics is only one. May be different
>> descriptions of entities satisfying to same results of experiments.
>
> Because according to the modern understanding of things, there cannot be
> experiments which could refute the uncertainty principle. In that sense
> any
> description must comply with the principle.
>
This is universal principle for clearly defined "intelligent system" and
with respect to complexity (or similarity of real world) of environment
(area) similar to that uncertainty principle in quantum physics depends on
speed of particles, size of particles (neutrino or nucleon or atom or
molecule and so on). Thus we may speak about uncertainty principle
relatevely in any circumstances. Same situation is in intelligent systems.
There are not exact bouder between stochastic and deterministic processes
and all our knowledges are relative.
I speak for my students that "I know just one absolute truth that absolute
truth is absent".
>>> Let it be this way (A). Let a DFA (NN is a case of) can think (B). Then
>>> A &
>>> B is in contradiction to your principle because a DFA can be described
>>> exhaustively in both its structure and behavior. Ergo, your principle is
>>> incompatible with computability of thinking. Good grief.
>>>
>> Could you implement DFA describing all our world with all details and
>> features? No, of cause. It is true for implementation of NN too.
>
> Ah, but only one intelligent thing is equivalent to the real world. That's
> Mr. God. I don't think that the objective of AI were to create gods...
> (:-)) Without falling into solipsism, intelligence is obviously less than
> the real world. Inside solipsism, intelligence as a concept is
> meaningless.
>
I don't understand or you don't understand me.
Meaning of my statement is that there is no any mechanism for full
description of universe (environment).
>> I don't agree that NN is case of DFA. Computer is DFA. But on this
>> platform
>> we simulate different processes which can not be described and
>> understanded
>> as DFA.
>
> As a DFA or by a DFA? So long NN is implemented by a DFA it is one. The
> question of understanding what and how such NN does, by some other system
> (possibly also a DFA) is another issue. If you didn't separate these, then
> indeed you would have a bad infinite recursion up to the whole real world
> and beyond.
>
If NN is implemented by DFA (or computer) you are right (but if we don't use
any generator of random values, not pseudo-random).
But even in this case if we have enough large number of inputs, outputs,
neurons and samples during learning (action) of NN it is impossible to
describe behavior of NN as DFA and this is nonsense.
And also deterministic behavior of NN is more disadvanage of it than
advantage.
If we implement NN in another platform (e.g., in chemical anisochronous
processes as in brain) we absolutely can not describe it by DFA.
>> Moreover, if we simulate thinking more far from determined processes
>> then we get more clear similarity of natural thinking.
>> And why you are sure that thinking must be computable?
>
> I am not. Probably randomness (free will, huh) plays a sufficient role.
> Maybe even stronger oracles do.
>
> But if you don't believe in its computability why do you bother with NN,
> AI
> etc? (:-))
>
Why any artist is working althogh he know that never he will be Leonardo Da
Vinchi?
It is intersting and important for him. Same for me.
Every step is step to achievment of perfection.
>> There are different opinions about it. See, for example, books and papers
>> of
>> Roger Penrouse.
>
>> Our perception of world is uncertain (fuzzy) and it is impossible to
>> demand
>> from our mind (or from good artificial intelligence) deterministic
>> decisions
>> (or responses).
>
> Well, if uncertainty propagates through the system, down to the actuators,
> we have to "defuzzify" it before firing actions. Whether defuzzification
> need to be stochastic or deterministic is a question. You could try
> alternatives randomly or use a pattern. There are cases where one strategy
> outperforms another. I don't think this can serve as an argument
> for/against computability of intelligence.
>
Of cause, actions are deterministic. About it another principle in my paper
which was source of this discussion:
"The principle of unity in fuzzy reasoning and certain other operations"
The basis of reasoning lies in operating with fuzzy images by means of a
process of associative recall of images. At the end of the process, a choice
of certain operations is carried out (recalling of it): it is therefore
possible to associate the successful choice (the solved task) with the
focusing of attention, the start of operation as programs of operation motor
neurons, etc. Thus the selected operation is involved as a tag in the
further process of reasoning.
Our fuzzy perception is converting to deterministic actions (or plan). May
be this property of animal's brain is reason of appearance of language and
abstract thinking.
In beginning it was language of actions (language of body), later especial
actions (connecting with generation of sound) were used for appearance of
speech and signs for verbal (logical) thinking.
> --
> Regards,
> Dmitry A. Kazakov
> http://www.dmitry-kazakov.de
>
Best regards,
Andrey Gavrilov
andr_gavrilov@yahoo.com
[ comp.ai is moderated ... your article may take a while to appear. ]
--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)
|
