Showing posts with label Theory of Intelligence. Show all posts
Showing posts with label Theory of Intelligence. Show all posts

Monday, November 27, 2023

// // Leave a Comment

On Understanding and Calculation, Quantitative and Qualitative Reasoning: Where Calculation begins Comprehension Ceases: Part II

 Comment on:

"Dr. Jeffrey Funk • 2nd•  Technology Consultant

Nobel Laureate Richard Feynman understood the difference between knowing and understanding, which is explained in this five-minute video. Knowing is being able to do calculations that agree with experiments. Understanding is being able to explain the underlying phenomena.

 

As Feynman describes, the Mayans knew positions of the moon and could predict eclipses, but they didn’t understand the reasons for their correct calculations. That understanding did not come until Newton and others explained gravity and its impact on rotating bodies. And the lack of understanding allowed the Mayans to falsely attribute things to gods, and not to physical laws.

 

Many data scientists and other proponents of AI point to knowing, being able to do calculations. Their #algorithms can predict the prices of homes and infections in hospitals, and match applicants with jobs or a camera’s input with a standard image. But they do not understand the why of their calculations or predictions. And when the calculations and predictions are wrong, they don’t know why. And unless they also have an understanding, which some call explainable #AI, the systems may always perform badly. Achieving high precision predictions or matching two things will always require understanding, not just knowing."

https://www.linkedin.com/posts/dr-jeffrey-funk-a979435_algorithms-ai-technology-activity-7133763690717224960-fMbK?utm_source=share&utm_medium=member_desktop

....

 

Todor Arnaudov - Tosh/Twenkid:

This is very similar to Arthur Schopenhauer, early 19th century: "When calculation begins, comprehension ceases.", OTFFR... His PhD thesis more than 200 years ago. A more modern term is "grounding" or "sensory-motor grounding".


" To calculate therefore, is not to understand, and,

in itself, calculation conveys no comprehension of things.

Calculation deals exclusively with abstract conceptions of

magnitudes, whose mutual relations it determines. By it

we never attain the slightest comprehension of a physical

process, for this requires intuitive comprehension of

space-relations, by means of which causes take effect."



See more in the part I from 2014: Where calculations begin, comprehension ceases* - on understanding and superintelligent AGI. Todor's comment in "Strong Artificial Intelligence" at Google+ https://artificial-mind.blogspot.com/2014/08/where-calculations-begin-comprehension.html


Chomsky's views on DL and LLM is similar as well, see MLST Youtube channel's episode "Ghost in the machine", 2023 etc. or even the "classical debate" with Peter Norvig from 2011, "Norvig vs Chomsky", "The Norvig - Chomsky debate": https://norvig.com/chomsky.html


However "explanation" and "understanding" are not explained in Feynman's wordplay either. What is to explain and explain to whom? If the receptor is not "appropriate" to your explanations, you will always fail. (See example videos on Youtube of "Explaining a concept at different levels: a child, a teenager, a college student, a graduate...", such as: Theoretical Physicist Brian Greene Explains Time in 5 Levels of Difficulty | WIRED: https://www.youtube.com/watch?v=TAhbFRMURtg ... The calculation models also "explain" and they eventually map to something different than the plain numbers or "abstract quantities" (unless the evaluator only reads the numbers), but they are too "shallow" (for a decided measurement) or the "reader" doesn't... understand them, she can't make some "other" predictions or make some new conclusions - all that she has expected that she should be able to do "if she did understand", "if it was well explained" - and she can't connect the evidence on her own - as she expected she should be able to do. Even if you "just" calculate the trajectories of the planets, without knowing "the reason", it eventually maps to objects known as "planets" and there are some additional conclusions such as predicting that particular stellar bodies will not collide etc. DL models may "say": "this is so, because: see this chain of vectors, 0.34, 0.23, 0.55, 0.343 ... at level 0.34, 0.344, 0.33 ... it is bigger than ... etc. deconvolve here, convolve here with this data point (they lose the records/path of thought of learning, IMO a better cognitive system remembers more details about the path and the mapping and should/could be able to find it).

The "calculators" response: "your "explanations" are just Qualitative generalities; no Quantities - No science!"

There is not so sharp distinction though, because the "calculation"-(quantitative-reasoning)-lovers may object, that the other ones' "explanations", if lacking calculation/math part, are "just qualitative", and qualitative-only theories or positions/statements etc. are supposed to have lower "resolution"/explicitness and usually are/may be not practical or can't be applied in specific scientific methods. They are just "methodological", but not "methods" as explained in one Bulgarian philosophical article by Sava Petrov*.


There should be *both* qualitative and quantitative models and representations, and reasoning is also a kind of "calculation", but the mappings to the "real" data is supposed to be preserved somewhere in order to connect to the "real" world (and to "explain" to the "user" sensory-input compatible modalities).


As the aforementioned physicist Richard Feynman also explains once in a BBC documentary, when he digs into the answer of "Why ice is slippery"(?), the "Why" questions can go deeper and deeper. There could be different depth and precision of "understanding" (or mapping to something else). Also the causality in general can go wider and wider in time, space and resolution up to the whole Universe and its whole universe. There is a limit, e.g. in Karl Friston's FEP/AIF it's "Markov blankets" which are taken as "impenetrable". There's a cut, a resolution of causality-control where the limit is set, and a model/expectations/causes which are accepted as "good enough" and somebody, the observer-evaluator decides, accepts, that "this is explained" or "understood", why when there is less depth, less steps, smaller range etc.: it is considered "not explained". See "Theory of Universe and Mind" by Todor Arnaudov, early works 2001-2004. One particular work with a lot of topics: "Analysis of the meaning of a sentence, based on the knowledge base of an operational thinking machine. Reflections about the meaning and artificial intelligence (...)", 2004  Translated in English in 2010 in 4 parts. https://artificial-mind.blogspot.com/2010/01/semantic-analysis-of-sentence.html 
https://artificial-mind.blogspot.com/2010/02/causes-and-reasons-for-any-particular.html  https://artificial-mind.blogspot.com/2010/02/motivation-is-dependent-on-local-and.html https://artificial-mind.blogspot.com/2010/02/intelligence-search-for-biggest.html TOUM continues with "Universe and Mind 6" (not published yet).

Compare "Analysis.." and TOUM with the later published theories and ideas, practical work as well (in RL, lately Active Inference, see the post about "Genius" platform from Verses AI), which confirm the claims and reasoning of the former, see:

https://github.com/Twenkid/Theory-of-Universe-and-Mind/


Read More

Thursday, June 24, 2010

// // Leave a Comment

Teenage Theory of Mind and Universe (Excerpts, Part 1) - Theoretical Induction of Neocortex and Hippocampus using Consciousness; Compression ...

If you've followed this blog long enough, you've certainly has read about my mysterious theory.  :))  It's a collection of writings and ideas on philosophy, mind and universe from my late teenage years (Late 2001 - Early 2004).

For patient Bulgarian speakers the originals are still available here:
Български:  http://eim.hit.bg/razum    (Съветвам да четете отдолу (най-стари) - нагоре)Откъси от "Тийнейджърската теория на разума и Вселената" - сбирка от статии, разработки, "трактати" от края на 2001-ва до началото на 2004.  Можете  да видите  откъсите по-долу, но новите допълнителни коментари са само към преведената на английски част, ако знаете английски, предлгедайте и тях.

Some of the topics in these excerpts are:

- Discussion on memory and consciousness
- Theoretical induction of neocortical and hippocampal-like modules in mind, based on consciousness. These modules are called respectively "Executive Operating System (EXOS)/Executive Memory" and "Events Operating System (EOS)/Events Memory"
- Compression and Predictability
- Mind as predictor of future inputs, based on the past
- Cognitive hierarchy and abstraction in mind - higher levels work with ever less amount of information, compress
- Art and compressibility (seems related to Schmidhuber's theories about curiosity)
- Find for yourself
- Have added some new comments on the topics.

Enjoy!

Notes: In square brackets [blah-blah] are comments/clarifications added now, some words and concepts are translated with many words/variants.

Man and Thinking Machine (2001)


"....In human [mind], most of the objects (of any kind) are not remembered "photographically", but are told/narrated in brain; only the most specific features of the informational entities are recorded; the input is compressed. Probably only the basic concepts are stored in "photographic", "phonographic" (audio), "text", "stereographical" (spatial) etc.  memory. (...) 

It's smart the newly arrived input informational entity to be interpreted by the existing concepts, using their labels - links to their meaning,  containing only an "address". We, people, call this kind of memorizing "understanding" and "making sense of" ...." - "


"Informational entity" is a portion of input data, that is analyzed/compressed into patterns.

...

Sections from:

Concept about the Universal Predetermination, a.k.a. "Theory of Universe and Mind", Part 3 (2003)

61. The origination of consciousness is the creation of the first advanced/well-developed/robust ("развит") Operating System for Events - EOS (in Slavic terms it was: събитиен вседържец). EOS stores and processes sequential information from the senses and ensures recording and replaying events (happenings) as on a tape or a disk - a record can be selected and it can be "played-back" in the imagination. 


The first well-deveoped operating system - EOS - is an OS, capable to keep compatibility with the previous versions, while it's being improved over time.

That means, the newer versions of the "formats" for recording memories are backward-compatible with the older ones, so older still can be read, even though the new versions may have extensions.

For example a person may have memories both as a two-year old and from yesterday - our brain has changed through the years, but it apparently has kept its compatibility with the records as a two-year old.

In memory, the old events/happenings (събития) - inputs to memory/records - are one-dimensional, i.e. all can be recalled in an instant. [Associative memory]

If a person is unable to spot anything particular - written, drawn; marked [to recognize, to spot anything recognizable/interpretable, to find a pattern], it's hard to remember it, because he would be supposed to record it as an image, while human brain is weak in remembering single complex images, not connected into a system.

After a person learns to read (has extended the capabilities of the EOS), when text is seen mind remembers not the image, but the letters, the syllables or even whole words. (Which maybe are represented as addresses in memory.) The capability to remember events/happenings (memories), written as text, is an upgrade/extension of the EOS.

However it's also an upgrade of the Executive OS [EXOS], which is more closely related to the "hardware". [More closely related to hardware means lower levels of control/generality.]

The memory for events is a higher level of memory in mind than the executive/action memory. For instance, executive memory is language - a baby says his or her first words usually when 1 year old. Before reaching to the moment after which he or she has memories kept for life, the baby might be using hundreds of words and even construct simple sentences. These words and grammars are [seem to be] a part of the Executive memory of the EXOS.

The abilities to speak; to coordinate his movements - walking, running, sitting, catching; recognize direction of sound; recognize objects; suppose third dimension data from a 2D image etc. are a part of a lower level of the OS of mind and apparently EXOS reaches to a backward-compatible versions much earlier than the EOS.

The first memories for events [biographical memories] display approximately when the first steady-working version of the EOS was finished, usually around 3 years [sometimes between two and three]. Earlier appearance display earlier brain development and this is an advantage for the further development of mind.

[It's related to the maturation of hippocampus, isn't it?]

EOS is more versatile (universal) than the EXOS, because it allows working with more data formats, which are combinations and extensions of the formats of the EXOS.

EOS is more flexible than the EXOS and allows mind to work with more abstract files ("informational entities"; records; patterns), more distant/abstracted than their form as an input. ["Distant" either as amount of transformations from the input to the record, and in space and time]

EOS allows mind to extract executive information from the memory for events, i.e. mind may interpret (reflect) EOS with much bigger delay in time.

Mind can convert memories for events into executive, in a format appropriate for the EXOS, thus EXOS may do particular actions because of events (happening/story/record) stored in the memory for events.

EOS can be seen, to a greater extent than EXOS as a data storage, which however eventually allows inducing sequences of actions/operations - algorithms [to be fed to Executive OS].

On the other hand, EXOS can be seen to a greater extent than EOS as an algorithmic storage. Executive OS uses Executive memory to execute actions. [Memory should be where patterns are stored, "operating system" is the process of activation of particular patterns.]

EXOS and EOS are interconnected and not independent. Features of any of them are features of the others, from particular points of view.

EOS is derived from EXOS.


# END-OF-SECTION


A speculation from now:

- Besides biological imperfections, I guess neuroplasticity may decline in order to preserve mind from "self-amnesia".

There are feedbacks between levels of generality, so they're supposed to adjust in parallel, generalizations should be reinforced/re-checked/re-affirmed; but I wonder what's going to happen with biographical memories, stored using hippocampus and "addresses in memory" (connections).

I've read that hippocampus eventually strengthens networks in neocortex, "transfers memory" there. However, if neocortex changes radically at the lowest levels of generality (or any level?), say if brain is exposed to adapt to a totally different world, I guess those connections may turn not just menaningless, but crazy. On the other hand, addition of higher or parallel levels of generality would not hurt the lower ones.

That's why I suspect that too high neuroplasticity in the lowest levels of generality after higher levels and hippocampus are developed may be dangerous and lead to insanity.

Also, brain is too messy to do real self-rewrite as a future thinking machine would be able to. Brain can't start building an improved "clean copy" and it can't isolate portions of itself from the others.

- Prediction of a "neocortex"-like module - Executive Operating System (EXOS) and executive memory (should be the patterns in the levels of generality)

- Prediction of a "hippocampus"-like module - Events Operating System (EOS)



62. The highest parts of the thinking machine [e.g. consciousness] work at the slowest speed and they output the least amount of information. The lower parts of the thinking machine, and of any machine, work at a higher speed, expressed as information flow/transfer rate.

Full control of a unit over another requires the first to contain in its memory a complete model of the controlled one - a complete model of its behavior - and the contol unit must be capable to simulate it [at least] in real time (...).

"Higher" parts of mind have narrower transfer rates than the "lower", that's why higher parts cannot actually have full control of the lower ones.


[Lower amount of information at higher cognitive levels - higher generality, ever reduced resolution/precision]

63. The more complex the unit/device*, i.e. more information is being transferred [between its parts] during its work, the less controllable it becomes, because it gets harder to find another device possessing enough memory and speed; the "devices" [agents might be a better term] with non-incrementing transfer rate (e.g. humans) may control the more complex devices the less, because their speed and memory gets lesser part of the requirements.

[There were other types of complexity used in the writings also, like one similar to minimal message length/algorithmic complexity, but not precisely defined. E.g. the length of a program written in the "machine code of Universe" which are capable to produce the entity whose complexity is measured, with two options - using or not using external sources for complexity. "Virtual universes" also exist, all environments having "laws of physics" and "matter" to evolve following the laws.
E.g. in this definition of complexity, even very simple artifacts created by humans are more complex than humans, because they require operational humans with developed intelligence, and cannot be created using the laws of physics without intelligent agents. ]

I think the "control" parts might have been partially aimed to show consciousness is just a top of an iceberg of mind, and to shut "Chinese Room" mouths and of those believing that computers do "what we tell them how to do".  What we consciously and voluntary do is pressing a few buttons, our holy free will is encoded in a few bits per second, while the vast accumulated complexity is doing the rest on its own, including suggesting and allowing "us" to press those buttons.

64. People wish for "richness" of the information they perceive, even though huge amounts of that "informational fortune" where mind lives are wasted.
[I.e. mind compresses, generalizes, doesn't use all data it gets.]

Richer in information a "file" [original - свитък, a piece of information, a message] is an entity/whole [a pattern], where predictability of the content of a piece of the pattern, based on other adjacent or related in particular fashion to the first parts of the pattern is close to the predictability in the Reality, in the First Virtual (Imaginary) Universe.

The mind concept (comprehension, understanding) that mind gets from an image, is the same, written as pieces of knowledge [patterns] and their relations.

Understanding shows what transformations can be done with the patterns, which create the file [input, a piece of input, a message...].

Abstract artists' pictures, where there a re no pieces of knowledge, except the most primitive for an image do not contain any meaningful information [or "information for mind", i.e. data that have patterns and can be compressed].
(Such primitive pieces/patterns are e.g. a point, spot, line, seasow line; polygons which mind does not associate with known objects; without 3-rd dimension).

There is meaningful information in images where mind can recognize and call particular objects, parts of objects and their coordinates/relations to each other.

The meaningful idea ("разумна представа") is described with much less information [data] than the image from which it was deduced. [Compression and selection.]

...

E.g. this image has a little knowledge, because it can be easily defined with a cycle in mind, where the changes of the bar of the building are small.
Photographers use to call such pictures "boring" - they bring a little amount of information, which is unpredictable using the precedent information [from the picture or whatever artifact being analyzed].  [This image is very highly compressible.]

Well done art photograph must be balanced: to bring enough information about the subject on the picture, to be clear enough - to be easy to understand what's on the picture, to have a "complete" idea; also, the picture shouldn't be overloaded with details which are distracting attention (are "irritating") and make our sight to roam around the details, searching for a meaning in the picture.

Art photography is based on abstraction and emphasizing of the essential [information]. It's like that also in classical visual art, especially in graphics and caricature.

[Aesthetics is not that simple to define in a few sentences and is subjective. There's "organic/smooth" patterns that seem to make perceptions generally more pleasant than highly compressible high contrast images; also there are high level associations, that pictures bring to us and may make us judge them emotionally etc.; "animate" objects and conflicts like in drama may engage mind to think/predict (interestingness)...]



Dogfight

Lonelyness (The Lonely Station Cat) |  Photos: Tosh (me)

A person creating any kind of art is displaying how the image of what he's creating is stored in his memory and what features of reality are memorized. For example the earliest children drawings just use lines - therefore [perhaps] images are stored as "lines"; lines bring the most possible information, because they are abrupt change between the colour of the paper and the pencil or whatever tool. The youngest mind remembers [maybe implies: recognizes] only the very essential parts of the perceived reality.

Humans don't remember lots of the details in the images. Maybe that's why pictures with lots of details are "irritating". Human mind usually removes the most of the slight details, when it remembers the image as a description of pieces of knowledge, in order to make remembering easier.

It's easier to fill memory using a piece of information [a pattern] to be multiplied with a gradual change, known from the beginning; and by parts of the image to be able to predict, with a higher probability, what's on other parts of the image. [That's easier, than to hold a bitmap copy.]

Human hardly can remember and percept too "plentiful information servings" such as noise:

NOISE: A sequence of characters/symbols, where every following has equal probability to appear and the value of the following characters is fully unpredictable for the one who's trying to predict it [i.e. the *generator* may actually *know* all the symbols and they might be pseudo-random for him, but the *evaluator* doesn't know them and can't predict them].

Mind works with big amount of redundancy in information, in order to predict the future inputs by the past easier and with higher certainty.

#

Български (допълнителните коментари обаче са само към частта на английски! предлгедайте я, ако знаете английски) 


*** При човека, повечето обекти (от всякакъв вид) не се запомнят
"фотографски", а се "преразказват" в мозъка, записват се най-характерните
особености на информационните обекти, входната информация се компресира.
Във "фотографска", "фонографска", "текстографска","стереографска"
(пространствена) и пр. памет вероятно се съхраняват само основните понятия.
(...) Хитро е новопостъпилият информационен обект да се обясни с наличните
информационни обекти. Просто му се дава етикет, а същността му се описва с
известните понятия, като се използват техните етикети - връзки към
значението им, съдържащи само "адрес". Ние хората наричаме такова запомняне
"разбиране" и "осмисляне". - из Човекът и Мислещата машина, Тодор Арнаудов - Тош, 2001

Из "Схващане за всеобщата предопределеност 3" -  Тош, 2003
61. Възникването на съзнанието е създаването на първия развит събитиен вседържец (събитийна операционна система) на човека. Събитийният вседържец запомня и обработва поточна информация от сетивата и осигурява запомнянето и възпроизвеждането на събития като "на лента" или "плоча" - можем да изберем "запис" и да го "прослушаме" във въображението си.

"Първият развит събитиен вседържател" е операционна система, способна при усърършенстването и надграждането си да запазва СЪВМЕСТИМОСТТА с предходните версии. Например форматите за запис на събитийни спомени от "Проявление 1.0" на Развития събитиен вседържец на разума нататък са съвместими един с друг "отгоре надолу" - по-новите версии са съвместими с по-старите, техни разширения.

Например имаме спомени и от двегодишна възраст, и от вчера. Мозъкът ни се е променял през годините, но си е останал съвместим към спомените от двегодишна възраст.

В паметта ни всички събития, когато и да са били въведени в паметта, са разположени едноизмерно и си спомняме за тях мигновено.

Ако не можем да четем и видим нещо написано, е трудно да го запомним, защото трябва да го запомним като цяло изображение, а човешкият мозък е по-слаб в запомнянето на единични - несвързани в система - сложни изображения.

След като се научим да четем (разширим възможностите на Събитийния вседържец), като видим написано слово запомняме не изображението, а буквите, сричките и дори целите думи. (Които може би се представят като адреси в паметта). Способността да си спомняме събития, написани в слово, е надстройка на събитийната операционна система. Същевременно е надстройка и на Изпълнителната, която е в по-пряка връзка с "железарията".

Събитийната памет е по-високо ниво памет в mind, отколкото изпълнителната. Изпълнителна памет е например езикът - детето проговаря, средно, на възраст една година. Преди първите спомени за случки, които оставят за цял живот, то използва стотици думи и построява изречения. Тези думи и граматиките са част от "Изпълнителната памет" на "Изпълняващия вседържец".

Способността да говори; да съгласува движенията си, за да ходи, тича, сяда, хваща; познава посоката на звука; разпознава предмети, предполага за третото измерение от двуизмерни изображения и т.н. са част от по-ниското равнище на Вседържеца на mind - Изпълняващият вседържец (ИВ).
Той стига до получаване на съвместими отгоре надолу версии много по-рано от Събитийния вседържец.

Първите спомени за събития показват приблизително кога е била завършена работеща устойчиво разновидност на Развития събитиен вседържец (РСВ) - обикновено за около 2-3 години след раждането на човека. По-ранното появяване показва по-ранно развитие на мозъка и е предимство за развитието на mind.

Събитийният вседържец е по-универсален от Изпълнителния, защото позволява на mind да работи с по-голям брой формати на данните, които са съчетание и разширение на форматите на Изпълнителния вседържец.

Събитийният вседържец е по-гъвкав от Изпълнителния и позволява на mind да работи с по-отвлечени свитъци (информационни обекти), по-отдалечени от вида им при въвеждането в паметта през чувствениците.

РСВ позволява на mind да извлича изпълнителна информация от събитийната. Т.е. разумът може да тълкува (обмисля) събитийната памет с много голямо закъснение във времето. Разумът може да преобразува събитийната памет в изпълнителна, във формат за Изпълнителния вседържец, който може да извърши определени действия заради събитие, съхранявано в Събитийната памет.

Събитийната памет може да се разглежда, в по-голяма степен от изпълнителната, като хранилище за данни, от които чрез изчисления могат да се извличат последователности от действия - алгоритми.

Изпълнителната памет е, в по-голяма степен от събитийната, хранилище на алгоритми (казбеници), отколкото събитийната; както подсказва името й, за да се извършват действия, се използва Изпълнителната операционна система.

Изпълнителният и Събитийният вседържатели се преплитат и не са независими един от друг. Свойствата на единия са свойства, в определена степен и от определена гледна точка, и свойства на другия. Събитийната операционна система е производна на изпълнителната.



62. Разумът е частта от Мислещата машина, която работи с най-ниска скорост и извежда най-малко информация. По-нисшите части на Мислещата машина, и въобще на всяка машина, работят с по-висока скорост, изразена в информационен поток (въобречие).

Пълна власт на едно устройство над друго изисква управляващото устройство да съдържа в паметта си изцяло паметта на подчиненото, модел на действието му и да може да му подражава с пълна скорост (това са условията за пълноскоростна емулация на една изчислителна машина от друга), за да е способно да ръководи всяка подробност от случващото се на подчиненото.

"По-висшият разум" има по-тясно въобречие от "по-нисшия", затова е обречен да няма пълна власт над него.

63. Колкото по-сложно е устройството, т.е. колкото повече информация се обменя при работата му, толкова е по-неуправляемо, защото е по-трудно да се намери друго устройство с достатъчно обемна памет и бързодействие, в които да се вмести подчиненото; толкова повече устройствата с нерастяща пропускателна способност (напр. човеците) по-слабо го управляват, защото бързодействието и паметта им стават все по-малка част от бързодействието и паметта на устройствата с растяща пропускателна способност, памет и бързодействие.

Най-сложното устройство - цялата Вселена - не може да бъде управлявано от никое друго устройство.

Колкото по-точно устройството знае какво прави - в паметта, достъпна за най-висшите части на управляващата част, е записано с по-голяма точност "какво му е писано" - толкова е по-свободно и същевременно е толкова по-несвободно. По-свободно е, защото прави това, което то "иска", т.е. което е записано в неговата памет; несвободно е обаче според разбирането, че "свобода" означава да не знаеш какво искаш и да не знаеш какво ще поискаш в следващия момент.

64. Човекът милее за "богатство" на възприеманата информация, макар че от цялото "информационно богатство", в което плува човешкия разум, се пилеят огромни "средства".

Информационно по-богат свитък е такава цялост, в която предсказуемостта на съдържанието на откъс от свитъка, въз основа на други, съседни или свързани по определен начин с първите, части от същия свитък се доближава по стойност до предсказуемостта в Действителността, в Първата въображаема вселена.

Разумната представа (съзнанието, разбирането), която получаваме от едно изображение, е същото,описано с късове знание и отношения между тях. Разбирането показва какви преобразователни действия биха могли да се извършат с късовете знания, които се намират в свитъка.

Абстрактните картини, в които не се срещат късове знание, с изключение на най-първичните за изображение (точка, петно, отсечка, начупена линия; многоъгълници, които възприемащият не свързва с познати предмети; без трето измерение), не носят никаква разумна информация.

Разумна информация има в изображения, в които могат да се разпознаят и назоват отделни предмети и части от предмети и взаимното им разположение.

Разумната представа се описва с много по-малко обем информация, отколкото изображението, от което е изведена.
...

Това изображение, например, носи много малко знание, защото лесно може да се опише с цикъл (въртележка) в разума, като промените спрямо цикличното изменение на мотива "ивица от сграда" са малки.
Фотографите наричат такива снимки "скучни" - носещи малко количество непредсказуема, въз основа на предходната, информация.

Добре направената художествена снимка трябва да бъде уравновесена - да носи достатъчно информация за предмета на снимката, да е достатъчно ясна - лесно да се разбира какво е снимано, идеята да бъде "завършена"; не трябва да бъде претрупана с подробности, които отклоняват вниманието ("дразнят") и карат погледа да се лута из изображението, в търсене на смисъла на снимката.

Художествената фотография е свързана с отделяне и наблягане върху същественото. Така е при рисуването, особено при графиката и карикатурата.

Когато човек твори, показва по какъв начин представата за това, което твори, е записана в паметта му и кои особености на действителността се запомнят.

Например малките деца, които правят първите си рисунки, използват черти - следователно в паметта на човека изображенията се записват като черти; чертите носят най-много информация, защото са рязък преход между белия лист и цвета на писалото; най-младият разум запомня само най-съществените части от възприеманата действителност.

Човек не помни много подробности от образите. Може би и затова снимките с много подробности "дразнят".
Човекът обикновено изтрива по-голямата част от дребните подробности, когато запомня изображението като описание на късове знание, за да се опрости запомнянето.

По-лесно е да се запълни паметта с късче информация, която се размножава с постепенна, известна отначалото, промяна; и чрез част от образа да можем да предскажем, с по-голяма вероятност, какво има на друго място от него. Човекът трудно запомня и възприема "обилни информационни порции", каквато е шумът:

ШУМ - последователност от знаци, в която всеки следващ е равно вероятен и стойността на следващите знаци е напълно непредсказуема (за този, който се опитва да предсказва), въз основа на предходните.

Разумът използва голямо излишество на информацията, за да може по-лесно и сигурно да предсказва бъдещите входни данни, използвайки миналите.
Read More

Wednesday, February 10, 2010

// // Leave a Comment

Intelligence: A Search for the biggest cumulative reward for a given period ahead, based on a given model of the rewards. Reinforcement learning.

Analysis of the meaning of a sentence, based on the knowledge base of an operational thinking machine. Reflections about the meaning and artificial intelligence.
Part 4 of 4 - Comment #3
Part 1 (и български): Semantic analysis of a sentence. Reflections about the meaning of the meaning and the Artificial Intelligence
Part 2 (и български): Causes and reasons for human actions. Searching for causes. Whether higher or lower levels control. Control Units. Reinforcement learning.

Part 3 (и български): Motivation is dependent on local and specific stimuli, not general ones. Pleasure and displeasure as goal-state indicators. Reinforcement learning.
Part 4 : Intelligence: search for the biggest cumulative reward for a given period ahead, based on given model of the rewards. Reinforcement learning.
One of the milestones of my AGI research. I wrote this particular article and the comments in Bulgarian as a 19-year old freshman in Computer Science at Plovdiv University.
By Todor Arnaudov | 13 March 2004 @ 21:49 EET | 340 reads |
First published at bgit.net and the e-zine “Sacred Computer” in Bulgarian

Comment # 3 by Todor Arnaudov | 18 March 2004 @ 22:04 EET
(...)

I'm one of the "scientists" who assume that since anything in Universe does work, there is nothing mysterious in its operation. I think that every information process in Universe is formal and can be modeled by a Fon Noeman machine if it has enough memory and - because Universe is working as a computer; because the computer is what it is, perhaps because it is following a high and universal model, of the Universe. Why the essence of the computers is yet a CPU, memory, clock [time, synchronicity], input-output since the start - this is the simplest complete model of operation of the Universe.

[As a University freshman it was harsh to call myself a scientist seriously, I was more a philosopher. This quotes my theory about "The Universe Computer", i.e. Digital Physics]


Everything is "formal" and "external", i.e. it could be written with formulas. E.g. why anyone assumes that another one is intelligent, if not external. And what is "internally intelligent" - one that is working like a human? What does it mean - to be built from proteins?...

Actually, a person usually says that something is "formal" if he believes that he does understand the formulas and/or if he thinks that these formulas are intelligible or superficial - just changing symbols like in Searle's Chinese room - even if he himself is incapable to understand them.

E.g. if a thinking machine is built and after it is taught [from a seed intelligence, a core, like a human] and it gets as complex as me or you, so that it would be intractable for anyone, some humans would still say that it is formal, they do understand it and it is nothing special.

Why would they think so? Because they assume that they do understand how computers work - "1 and 0, NAND, NOR; cycles, reading, writing, shifting... it's so simple and the computer doesn't realize anything of what it computes, so even if he appears to think, it would just calculate!"

Does the carbon atom "realize" that it's a part of a neuron?! Or each neuron "realizes" that it's a part of the brain, and wouldn't it behave the same way if we put it outside of the brain and feed it with the same signals like in the brain? Is the neuron aware that it's a part of a mind and does it know a part of what thought it is?

Bullshit! One thing is for sure to me - humans want to feel unintelligible, mysterious, hyper-complex, because people believe that to be "formal" or to be explainable is a bad thing. People don't want to ask hard questions, and more likely - don't want to get irritating answers. (...)

However, what is to be a human? I would ask, perhaps as a "self-humiliating" individual.
A new born is, not a 20-30 years old. A seed intelligence where mind grows and develops.

The mysterious concept of soul is an example of a helpless "prove" that human is not a "formal" entity. It has a soul, actually it is just something that a machine or anything else couldn't have just because it is a machine, that means - not a human, that means - not a member of our precious club. However those people do not question why most people believe that the animals have one, but machines - don't, while a lot of pretty formal reasons could be found.
(See the Teenage Theory of Universe and Mind/Intelligence and the short novel "The Truth" http://research.twenkid.com )

To me human is formal, and the principles in his mind as information processes are similar to what they could be in a thinking machine: memory, processor (merged in one), clock (a way to separate/resolve events), input-output. In the memory there are records of possible actions from which the person - his body - all the time selects one that mostly matches current goals for a given selected period of time ahead, based on a current active model for checking whether the goal is reached or not.

The purpose is a GOAL; a state, where the Control Units aims to. The "thirst for a purpose/meaning" is a search for local maxima, indicating to the Control Unit how close the goal is. [E.g. - an orgasm... ;) ]

The machine, as of my current understanding [2/2004] - should be made to have complex enough subunits - diverse enough in their kinds and goals, and either real or virtual - which would feel "pleasure" in a particular way, i.e. they would seek achieving their goals. And it should be impossible all subunits to reach their goal in the same time.


Terminal devices are needed (effectors) - muscles, which can't be controlled in parallel, e.g. only one subuint could control them at a given moment, so that the external behaviour, movement, output data - be clearly defined, as with human.
The subunits should compete and interact; to contract and fight; sometimes to collaborate in groups and to "fight" against other groups, aiming at the goal each of them to reach maximum pleasure.


I think that's what human does and I think [as of 2/2004] this is "mind" [intelligence]: search for pleasure and avoidance of displeasure by complex enough entities for a given period ahead [prediction, planning]; the "enough-ness"" is defined by the examples we already call "intelligent": humans at different ages, with different IQ, different characters and ambitions.

The behaviour of each of them could be modeled as a search for the biggest sum of pleasure (displeasure is a negative in the sum) for a selected period of time ahead, which is also chosen at the moment of the computation.

"Happiness" is another word for pleasure, and as a friend of mine, a poet, once said:
"The man is restless happiness seeker. I don't know why it's so important for everybody to take their dose of happiness, but it is true... and the people, caught in the hands of vices and addictions are a kind of deluded seekers... They are also searching for happiness, but at a wrong place".

Addictions - such as to drugs, but not only - are an example of Control Units finding their goal - an enormous pleasure they feel, taking a given drug - and this can lead to taking control over the whole body. "Happiness" for the body is turning to an elementary happiness for the elementary control units, which feel that their goal is reached by detecting a particular kind of molecules.

PS. When I was about to write "a local maximum" I was about to say also "for a close enough neighborhood" (very short period), influenced by the Mathematical Analysis I've been studying lately, however mind is much more complex than the Analysis, i.e. it's based on much more linearly independent premises, that is impossible to be deduced from each other.

It's not required the period for the prediction to be short enough, because like in the example with the chocolate, caries and the dentist, immediate actions can be included in functions that give result shifted far ahead in the future; at the same time, the same action can be regarded in the immediate future, in the example situation, the eater will feel ultimate pleasure in the following one second.

One among many functions can be chosen or constructed, and they all can give both negative or positive results, depending on minor apparently random details - the mood of the mind in the particular moment; which variables are at hand, which the mind has recalled of, what the mind is thinking of right then...

(...)

Suggested reading - "The Truth", a short novel I wrote in 2002. [not translated as of 10/2/2010]



Comments from 10/2/2010:

Yes, multi-agent stuff reminds Minsky's "Society of Mind", but I hadn't read it then and still have not.
The drug stuff is related to dopamine, endorphins and other neurotransmitters, "Reward pathway" ans so on. The discussion is about "virtual" units, Control units , not about neurons etc., though.
Also, the last discussion is about Reinforcement learning bug, I didn't know the word "reinforcement learning" back then and didn't know what exactly behaviorism was, I knew about utilitarianism, though, and was using my imagination to find explanations.

A version of reinforcement learning was reinvented in my early
works from my Teenage Theory of Universe and Mind, where the Control Unit (agent/human/society/state/system) is predicting and aiming at scenarios which would lead to the biggest cumulative pleasure for a given period ahead.

In society and states, most of the laws have this function, and it's related to another concept - the limited level of pleasure (reward in standard RL terms) that any control unit/subunit/agent can get in any possible circumstances. This is supposed to prevent one from gaining too much control over the system - mind/society.
I gave the following example: when somebody robs a bank, he gets rich, can do better what he wants, but his happiness is not unlimited, it's normalized to 1 anyway. On the other hand, he is causing lasting displeasure to many people and the sum is highly negative. The system is aiming to avoid robberies. This is related also to doing what one wants and what doesn't want (another important concept in my theory about Control Units). Control Units are aiming to control, and losing control and predictability causes displeasure/confusion.
Recently I found that Marcus Hutter and Shane Legg talk about a concept which is similar to my limited reward "Assuming that environments return bounded sum rewards ... " in "Universal Intelligence:A Definition of Machine Intelligence". Here. They also define the Universally intelligent agent as a seeker of a sum of rewards. However, his papers are published years later (or around, I haven't checked all), and I didn't know about him at the time, I heard of him in 2009.

In 2002-2004 when I was building the foundations of my Universal AI theories and understanding, I was too young, busy and "ignorant" - I didn't know about any of the gurus such as Goertzel, Schmidhuber, Hutter, de Garis. Hawkins has just appeared with his "On Intelligence" right after I published the last 4-th part of my first works, which was written
mostly in 2003.

I wrote my stuff on my own
and I still believe that "Imagination is more important than knowledge"; can't be so ignorant as back then, though.
I wasn't sure whether I'm not a mad man at the time, because it's hard to find anybody to understand and appreciate such advanced stuff as AGI, I was 18-19 years old. Years later, starting from 2007, I started to find that my ideas, pieces of them or directions are shared by gurus in AGI - first Hawkins in his Memory-prediction framework and "On Intelligence", then Boris Kazachenko, Jurgen Schmidhuber, Marcus Hutter.

Of course, this "Teenage Theory of Universe and Mind" is something that will be translated and shared as well.

Keywords: Reinforcement learning, multi-agent systems, control units, pleasure seeking, utilitarianism, local maximum, Todor Arnaudov's Teenage Theory of Universe and Mind, Todor Arnaudov's Teenage Theory of Universe and Intelligence, limited reward, limited pleasure, bounded reward, bounded pleasure, addiction, addiction bug of reinforcement learning, cumulative reward, cumulative pleasure, pleasure seeking, seeker, searcher, intelligent agents, Twenkid Research
http://research.twenkid.com
Edit @ 30.12.2018: Corrections of typos: shifter, loosing, predicatability, utilitarism, cummulative, circumsances, displeausure, smiliar, bulding, enourmous, neighbourhood,amongst, construted, unresting, muscules, theTeenage, untracable, synchron, input-otput, endorfine ...
Edit @ 3.9.2025: "its ..." (of human) - his; and one other similar. Visit the virtual conference Self-Improving General Intelligence SIGI-2025 and find the "Prophets of the Thinking Machines": https://github.com/Twenkid/SIGI-2025/
Read More

Sunday, May 31, 2009

// // Leave a Comment

Todor Arnaudov's story in Artificial Intelligence - FameLab 2009 Final, Bulgaria


Famelab Competition, Bulgarian final, "Sofia" theatre in Sofia, 11/5/2009

My story in Artificial Intelligence, or The Building Block of Intelligence










My story: hierarchical temporal memory, Jeff Hawkins, Numenta, Artificial Intelligence, neurons; cortical column; Todor Arnaudov's theory of the Universe and Mind from the early 2000s.


It was great. So many new friends from different fields of science, and even few people from the Show business, and several future actors in my movies. :)
в мои филми. :)

Thanks again to Bozhidar and Lyubov and to my friends who came to support me live.

Bulgarian Winner - Aneta Yovcheva



In the gathering after the final, with Ani and two friends from the first stage in Plovdiv.



Read More

Friday, September 26, 2008

// // Leave a Comment

Theories of Intelligence and Mind by Todor Arnaudov, Jeff Hawkins, Boris Kazachenko and others - Teaser

Български - после...

Did you know, that there are few "mad scientists" around the globe, who are claiming they do have theories of intelligence, theories of mind or something? :)

Probably the number of researchers in that direction has grown, since the most famous amongst them, Jeff Hawkins has published a book about his speculations in 2004. Actually my theory, which was about Mind, Universe and everything, was conceived and constructed between 2002 and 2004, I was still a teenager.

However it is still published only in Bulgarian: http://eim.toshuniverse.com/razum ... :)
Translation is in progress, from time-to-time.

...

In essence, it seems intelligence is a simple thing...

All of us has reached to conclusion, that intelligence is an ability to predict future perceptions by simulating or extrapolating perceptions taken from the experience. (Well, actually a lot of researches has reached to something like this in Machine learning, but... Later... )

Also, all reach to concepts related to hierarchies for predictions or simulations. Jeff calls it "Hierarchical Temporal Memory". Boris uses levels. I use also levels. I like "Virtual sub-universes in different levels", because my theory was general, metaphysics, it was finding patterns in Mind and in Universe.


Implementation of this simple idea, detailing... Modeling the dynamics of the system or the "scaling" of the system, as Boris says - this is the tricky part.

I'll write more about the theories of me, Jeff and Hawkins later.


Others?

I know few other ingenious people, potentially or actually in that special field... One of them is a mysterious woman, let's call her T-X. She was very advanced as I knew her 4 years ago, if I remember correctly, she was claiming that she is able to do real simulations of evolving mind, but they are too slow. I haven't spoke to her since 4 years, and she does not have a blog... :)

Hugo De Garis

Evolving neural systems... http://iss.whu.edu.cn/degaris/

Raymond Kurzweil
An ingenious engineer in the past, but I haven't heard he was creating something interesting recently. Mostly doing "AI prophecies" in books, explaining the future of thinking machines...

It's funny that we do share the passion in sound-processing and speech-synthesis - he was creator of the first OCR-Text-to-speech system in 70-ies. But this is another story, and one of his companies is for Keyboards, Synthesizers.

BTW, I finally got one :P, but for now it's just a basic Yamaha...

Wish one day I play a Kurzweil's synthesizer...




But this is yet another story. :)))

And yes, he's making good synthesizers, but I'm disappointed by his Cybernetic Poet. Yet another shallow pseudo-creator, which shuffles words, but doesn't have images and lacks imagination.

Hans Moravec

He is an AI prophet, also... :))

Usually "mad AI-scientists" are also AI-prophets... :))

Kevin Warwick?

Well, I wouldn't put him inside this group. Too... Too cyborgial. But let's mention him...


...

Some of those "mad-scientists" claim they are transhumanists. Some - almost with religious feeling.


A transhumanist psalm

The future of humanity is in machine bodies.
We will evolve to a higher form of existеnce!
We will be able to live forever!!!Eternity shall be ours, with our transhumane cybernetic ultra-giga-terra-peta evolved souls!

Amen!

I'm not one of the religious ones, or at least I don't think so, even tought part of my prose or metaphysics, which I wrote in my teens about that, seems the opposite... :)

I don't know... It might be cool... ;)



...To be continued...
Read More