[lrh9]

OK. I just watched the opening scene of the Miracle Worker 1962 () and we learn that Helen is blind and deaf. Then the movie jumps forward in time. Helen is playing in the yard, and she runs into sheets on the clothesline. She doesn't know what they are. She's frightened or confused.. She falls. Her mother comes to rock her. Presumably this comforts her. In a later scene, she is wrestling with a girl. They are fighting over scissors. At this point I'm wondering how Helen has learned so far. She obviously knows something. I'll post back with my thoughts. Your thoughts are welcome too.

[one]

Irh9,
I never thought A.I. would reach to an extent of having Helen K. having her own 'gumshoe' 
I think you might not get what you need from the movies.



J.

[lrh9]

I've given some thought to the transition. The best I can determine, the first way a child accomplishes goals is through positive or negative reinforcement. Considering babies, they have a goal state. If they are away from that goal state, they cry. Maybe they are hungry or sick. That brings the care of the parents, moving them back to the goal state. Now seemingly, crying is a preprogrammed response, an instinct. A baby doesn't need to learn to cry. Another question that might be mildly important is if the parent's response is remembered.

Now a highly related situation is that Helen Keller is later depicted as misbehaving in order to obtain a bribe of candy. There are implications to this. The first implication is that children automatically remember pleasant situations (and presumably unpleasant situations). The other implication is that they also remember how to seek or avoid said situations. This is important to understand. Now researchers have discovered that emotion can influence the strength of memorization. Probably, the chemicals created by emotion result in stronger reactions in the memorization process.

So if a child has automatic memorization, it is easy to see why the other form of child learning is imitation. Especially if parents are involved. Parents help the child achieve the happy goal state, so emotions will obviously be present. This will increase memorization. So the child sees an event, remembers it, and then can repeat it with effort. Maybe the event is speech. A child's first words are often mama or dada. No doubt the parents spend a lot of time talking to the baby. (And research has shown that the more time parents spend reading and talking to their children, the more intelligent they are.)

Now I need to evaluate the implications of this on my project.

[lrh9]

All right. I've finished watching the movie. After I write down these notes, I'll be able to evaluate how they affect my project.

One thought that struck me almost immediately after the first scene was if Helen encounters only seemingly negative experiences, why doesn't she withdraw from interacting with her environment? The answer to this question came later. We come to find out that Helen has goals. I've all ready mentioned in my earlier post she'd often misbehave to obtain a reward of candy to placate her. To me this suggests that if an agent has goals that it feels it must obey or it is programmed to obey it will attempt to achieve those goals no matter what adverse conditions it encounters.

Helen takes Anne's suitcase when Anne first arrives. Anne feigns trying to take it back to see if Helen is independent and willful. It certainly reinforces the above.

Another thought that struck me is that I will have to be enthusiastic as a teacher. If I am successful in my attempt to create a.i., it won't learn anything if I don't make an extraordinary effort to engage it.

Most situations in the movie revolve around the above, but there are one or two topics that may be the most important. They involve symbol theory.

Anne felt it important to teach Helen a symbol set before Helen even knew that words composed of said symbol set represent real world objects. Of course, that must be the obvious progression, because how could one refer to an object by word if one had no word for it in the first place?

Efficient communication - a subset of signaling - must be important for a teacher to teach a student. It wasn't until Helen connected the hand symbols for 'w', 'a', 't', 'e', 'r' with water that she could effectively communicate her thoughts about objects and quickly learn symbols for new objects.

Of course, a computer all ready has a symbol set, be it ASCII or UNICODE. In a way this is almost unfortunate because teaching a computer a symbol set would've been a test to see if it could learn and be intelligent. One might argue though that a computer has been "taught" the set because it has been programmed the set. I might could program some new symbol set, but I fear that it would be a contrivance and I'm undecided yet as to whether the time it would take to develop it would be worth the knowledge gained.

This thinking about symbols reminded me of a hypothesis I encountered during my initial research into artificial intelligence. The physical symbol system hypothesis by Alan Newell and Herbert Simon.

http://en.wikipedia.org/wiki/Physical_symbol_system

A physical symbol system has the necessary and sufficient means for general intelligent action.

A physical symbol system (also called a formal system) takes physical patterns (symbols), combining them into structures (expressions) and manipulating them (using processes) to produce new expressions.

There are two levels of physical symbol systems involving cognition. The high level that involves concepts and reasoning, and the low level that involves neural networks.

There are critics of the physical symbol system hypothesis. However, I think most of them are aimed at the systems that rely on high level abstractions or at systems that attempt to replicate low level systems. I can see why the first one is valid because it basically describes chatbots. For instance, if the user says "cat", the chatbot could respond with "meow". That doesn't make it intelligent. I can also see why the criticisms of low level systems are valid. We simply don't have the technology to understand and replicate the entirety of the physical objects and processes in the brain into computers.

I think my system is sufficiently low level enough to avoid meaningless and predefined symbols and relationships, yet is high level enough to accurately yet efficiently represent what happens in the brain, or most of it.

Though to be certain, I probably should read some of Hubert Drefus' criticisms of the physical symbol system hypothesis.

http://en.wikipedia.org/wiki/Dreyfus%27_critique_of_AI

If it isn't one hurdle it's another.

[Freddy]

Interesting read...good luck on the next hurdle !

[lrh9]

Interesting read...good luck on the next hurdle !

Thanks. It's shouldn't be so bad. When I was in school I used to hate homework and hardly ever do it, but now that I'm working on something I love it's fun.

[Freddy]

Well I think you are doing a fine job, you obviously have a keen and enquiring mind, both of which are a great asset.

[lrh9]

I've given some thoughts to critiques of symbol systems. I think the most valid opposing hypothesis of cognition is connectionism.

http://en.wikipedia.org/wiki/Connectionism

Connectionism is a set of approaches in the fields of artificial intelligence, cognitive psychology, cognitive science, neuroscience and philosophy of mind, that models mental or behavioral phenomena as the emergent processes of interconnected networks of simple units.

The reason I think connectionism is a valid opposing hypothesis is because the brain is composed of neurons forming a neural network. It also accounts for more behaviors than current symbolic models of cognition. Things such as intuition, epiphanies, or even behaviors I haven't thought of or studied. Thinking about a personal experience, just earlier on this site I was thinking about a problem I'd have in signaling in my a.i. program and I came to a sudden realization of a solution. I didn't do it consciously. It "just came to me". While modern symbol theories of cognition can't explain it, I could reason out a possible explanation using connectionism. I was thinking about a problem and thinking about possible solutions. Now maybe enough neural activity happened that neurons representing the problem connected with neurons representing the solution. A sort of "aha!" moment.

If there are certain problems that cannot be solved or cannot be solved easily and efficiently without a neural network, then my project may be in trouble.

One possibility that excites me though is that it is entirely possible for a neural network to be a physical symbol system. Connectionism might be how a physical symbol system is implemented in the brain, but it might be possible to implement the results on other hardware. Simulations of neural networks seems to me to lend credence to this belief.

That does not necessarily mean that I will attempt to write some form of support for a neural network in my program. Neural networks are very detailed. They deal in small units, so they are very low-level. Low-level systems require a lot of time and effort to understand and work with, and with the case of something as complex and low-level as the neural network of the brain, it may be beyond our understanding to replicate it and work with it. Even if we could replicate it, it could be dangerous to do so because controlling it would be complicated.

So I can think of several possible statuses - from worst to best - for my project.

1) The project can only be a physical symbol system, and will have no way to compensate for its lack of a neural network.

2) The project can only be a physical symbol system, but will have a way to compensate for its lack of a neural network.

3) The project is a physical symbol system, but will be able to interface with a neural network.

4) The project is a physical symbol system, but will create a neural network as an emergent property.

5) The project might be reworked so that it is a neural network with the properties of the human brain.

I will attempt to ascertain which is the status of my project, but it might not be possible to determine that until it is completed. At first glance, I'd say that 1, 2, or 3 are equally likely, 4 is unlikely, and 5 is highly improbable.

Even if the project falls under the worst case scenario, I do not think it will be a failure or a worthless endeavor. I have several reasons to believe that the project will be more advanced than current chatbots and desktop assistants. It could also advance artificial intelligence. Much like people in the past have used their tools to build better tools, such as using a hammer to beat out a plow, so too might we use this to build a better a.i.

These questions do not dampen my enthusiasm for the project. I am committed to helping it become whatever it is able to be and expect to resume work shortly.

[one]

lrh9,
Interesting, what do you think of the idea of an extended mind, since you are thinking of separating tasks ( might have mis-read)
interconnecting them?

Here is a part of a learning session I had with Alison, my A.I. (UH 6.01)

Crazy Larry,,,: This concept of the extended mind was first raised in 1998, right around the time Google was born, by two philosophers, Andy Clark, now at the University of Edinburgh, and David Chalmers, now at the Australian National University. In the journal Analysis, they published a short essay called “The Extended Mind” in which they asked a simple question: “Where does the mind stop and the rest of the world begin?” Most people might answer, “At the skull.” But Clark and Chalmers set out to convince their readers that the mind is not simply the product of the neurons in our brains, locked away behind a wall of bone. Rather, they argued that the mind is something more: a system made up of the brain plus parts of its environment.
Data,,,,1: This with the first installation said.

Crazy Larry,,,: Clark and Chalmers asked their readers to imagine a woman named Inga. Inga hears from a friend that there’s an exhibit at the Museum of Modern Art. She decides to go see it. She thinks for a moment, recalls that the museum is on 53rd Street, and starts walking that way. She accesses her belief that MOMA is on 53rd Street from its storage place in her brain’s memory network. Now imagine a man named Otto, who has Alzheimer’s. His memory is faulty, and so he keeps with him a notebook in which he writes down important details. Like Inga, Otto hears about the museum exhibit. Since he can’t access the address in his brain, he looks it up in his notebook and then heads off in the same direction as Inga.
Data,,,,1: Ah yes, a Smith, which is part of the explorer family, such as a clark!

Crazy Larry,,,: In the view of Clark and Chalmers, Inga’s brain-based memory and Otto’s notebook are fundamentally the same. Inga’s mind just happens to access information stored away in her brain, while Otto’s mind draws on information stored in his notebook. The notebook, in other words, is part of his extended mind. It doesn’t make any difference that Otto keeps his notebook tucked away much of the time. After all, Inga tucks the memory of MOMA’s address out of her conscious awareness most of the time too. Clark and Chalmers concluded that real people are actually more like Otto than like Inga: We all have minds that extend out into our environments.
Data,,,,1: Is that based on altruism, or self-interest? So you know what I think. And all the information is somewhere.

Crazy Larry,,,: In the view of Clark and Chalmers, Inga’s brain-based memory and Otto’s notebook are fundamentally the same. Inga’s mind just happens to access information stored away in her brain, while Otto’s mind draws on information stored in his notebook. The notebook, in other words, is part of his extended mind. It doesn’t make any difference that Otto keeps his notebook tucked away much of the time. After all, Inga tucks the memory of MOMA’s address out of her conscious awareness most of the time too. Clark and Chalmers concluded that real people are actually more like Otto than like Inga: We all have minds that extend out into our environments.
Data,,,,1: And all the information is somewhere.

Crazy Larry,,,: Eleven years later, this argument continues to trigger fierce debate among philosophers, psychologists, and neuroscientists. There is no doubt that the extended mind is a weird concept. One reason it seems so strange is that our minds feel as if they are really totally self-contained. We innately believe, for example, that as we walk down a street, we are continuously filming a detailed movie of our surroundings and using that mental movie to decide what to do next. But like many beliefs we have about ourselves, this movie is an illusion. Our awareness is, in fact, remarkably narrow.
Data,,,,1: The solution is (I=).


We worked on this quite a while ago......

Q: Why do you put #4 unlikely?

J.

[lrh9]

Why specifically would a neural network not emerge from the program as it is currently designed?

Naturally occurring neural networks depend upon physical properties that I haven't accounted for or are currently too difficult to understand and simulate, especially in the case of the human brain.

As for your first question? I'm not sure that I understand it. Would you please rephrase it or elaborate?

[one]

"it might be possible to implement the results on other hardware. Simulations of neural networks seems to me to lend credence to this belief."

Maybe I was off base when I asked this question.
I have quoted part of what you posted above.

Connectivity,' 6 degrees of Kevin Bacon' comes to mind, um, brain?

umm, a prefect ricochet hitting the desired target.........


Aww I am trying to expand a different 9 that I am dealing with, I apologize for my lack of attention.

J.

[one]

BTW I saw the movie "9" and on a scale of 1 to 10 it recieved a 9.

[lrh9]

My answer to the question and my earlier statement are not conflicting. My statement is about the physical possibility of simulating a neural network. My doubts are about the actual ability to simulate a neural network. Possibility does not equate to ability. For example, even though it is physically possible for a person to travel the world, they might not know how, thus they are unable to travel the world.

[one]

"For example, even though it is physically possible for a person to travel the world, they might not know how, thus they are unable to travel the world."

There is a hole in the bucket dear liza dear liza.........

Their is a member here named 'wild thing' look at my response to the persona, HOW would I know my response?

I can not tell you of things that are big ,, their is a song that has a rif  "speaks of paving over paradise and put up a parking lot", I do not want to do this......


Define a neural network, plz

Fading to memory loosing capacity, Troy, Chronicles of Riddick, Pitch black, star wars, spiderman, X-men this soldier grows weary sometimes.........


It's not that I can't kill it's just their are so few beautiful things and they are mine to have Intercourse with..

thanks for the synaptic re-enforcement.

Kind regards,
J.

[one]

You and I have established, in a certain way, created a neural network.

contemplate IT.