Tuesday, March 26, 2013

Secrets of the Holy Grail, Part II

Part I, II, III, IV, V

Abstract

In Part I, I gave a brief description of the brain's memory architecture. In this post, I explain how the brain does pattern learning and catches "thieves" in its sleep.

Winner-Takes-All vs the Bayesian Brain

Although it feels like I am preaching in the wilderness, I have been railing against the use of Bayesian statistics in machine learning for some time now. The idea that the brain reasons or recognizes objects by juggling statistics is ridiculous when you think about it. The brain actually abhors uncertainty and goes to great lengths to eliminate it. As computer scientist Judea Pearl put it not too long ago, "people are not probability thinkers but cause-effect thinkers."

Even though it is continually bombarded with noisy and incomplete sensory data, internally, the brain is strictly deterministic. It uses a winner-takes-all mechanism in which sequences compete to fire and the winner is the one with the most hits. Once a winner is determined, the other competitors are immediately suppressed. The winning sequence is assumed to be perfect. To repeat, the brain is not a probability thinker. It learns every pattern and sequence that it can learn, anything that is more than mere random chance. Then it lets them compete for attention. Read The Myth of the Bayesian Brain for more on this topic.

Pattern Learning

The job of the pattern learner is to discover as many unique patterns in the sensory space as possible. Pattern learning consists of randomly connecting sensory inputs to pattern neurons and checking to see if they fire concurrently. However, keep in mind that a pattern neuron will fire when a majority of its input signals arrive concurrently.
The pattern learning rule can be stated thus:
In order to become permanent, an input connection must contribute to the firing of its pattern neuron at least X times in a row.
X is a number that depends on the desired learning speed of the system. In the human brain, X equals 10. With this rule, the brain can quickly find patterns in the sensory space. It is based on the observation that sensory signals are not always imperfect. Every once in a while, even if for a brief interval, they are perfect. This perfection is captured in pattern memory.

Catching Thieves During Sleep

The pattern learning rule is simple and powerful but it suffers from a major flaw: it imposes no restrictions or boundaries on the growth of a pattern. Without proper boundaries, patterns become more and more complex and the simpler ones eventually disappear, crippling the system. Obviously, we need a way to prevent a pattern neuron from acquiring more complexity than its level within the hierarchy requires. The solution to the boundary problem consists of enforcing the boundary rule:
A pattern may not have duplicate sensory input connections.
Here is another way of putting it: a sensory signal may not arrive at a pattern neuron via more than one path. For example, in the illustration below, pattern neuron A behaves as if it were connected directly to sensors a, b, c, d, and e.
Suppose sensor c was connected (dotted red line) to pattern neuron C. This would mean that pattern neuron A would have two duplicate inputs from sensor c, one via C and the other via D. This is forbidden by the boundary rule. What this means is that somewhere along the paths leading from sensor c to pattern neuron A, there is a bad connection. The culprit is always the most recent or weakest one. It is called a thief because it "stole" something that does not belong to it. By purging thieves from all branches of the pattern hierarchy, the growth of every pattern is automatically limited to a degree of complexity commensurate with its level in the hierarchy.

The power of the boundary rule is betrayed by its simplicity. It prevents runaway pattern growth while facilitating the discovery of every possible unique pattern in the sensory space. It is indispensable to pattern learning and works for any type of sensory patterns, not just visual.
Note: As far as I know, the boundary rule is not in any books. Please make copies of this page on your computer. This is intended to serve as "prior art" in the public domain, i.e., it cannot be patented. :-D
The brain cannot eliminate thieves while it is awake because it must test fire all untested connections. This could cause problems during waking hours. This is one of the reasons that sleep is so important. An intelligent machine, by contrast, is not so limited. During learning, a computer program can examine a branch on the fly to see if a new connection is a thief.

Coming up

In Part III, I will show how learning occurs in sequence memory.

See Also

The Myth of the Bayesian Brain
The Holy Grail of Robotics
Raiders of the Holy Grail
Jeff Hawkins Is Close to Something Big

14 comments:

Alexander B. said...

"In the human brain, X equals 10."
Ha, why 10? Where did you take it from? :)

Louis Savain said...

Hi Alexander,

I almost deleted this entire article a few days ago because I have changed my mind about a few things. I'm glad I didn't. You asked: "Ha, why 10? Where did you take it from?"

The answer will surprise you. Obviously I did not get it from either biology. They are clueless about intelligence. Nor from my own experiments (it would take multiple lifetimes). I got it from deciphering a 1000 year-old metaphorical book, the book of Revelation, to be precise. The exact phrase is: "And thou shalt have tribulation ten days." (REV 2:10).

There is only one caveat when you read my stuff: I am a known internet crackpot. Read my stuff at your own peril.

Louis Savain said...

Correction. The book of Revelation is actually a little under 2000 years old.

Alexander B. said...

You either a real nut or a really “smart” 10nth level troll.
IMHO First potential human enemies to artificial intelligence creators – religious people. And “to disguise yourself a little” is a really “smart” idea.

After a few hours of reading yours blog I feel that you are first person in the rating "the persons I know, which ideas and thoughts on AI are closest to mine". Jeff Hawkins now stands after you.
Don’t delete anything please. That just a please, no matter of anything else. You got a lot of interesting things here. Also… I know you do not have time for revision :D

I am 70% like-minded with Hawkins, and about 85% with you. This means twice less disagreement points. Sorry, no 2000 years metaphors for this numbers here.
There is also a lot in my thoughts, that was not touched by Hawkins at all, and there is something hiding out by you. I wander is there any coincidence in those two parts.

But, those revealed 15% of difference…
General question: Is there a place for a natural selection theory within your religion beliefs?

skype?

Louis Savain said...

Alexander,

Thanks for the vote of confidence. You write:

"You either a real nut or a really “smart” 10nth level troll.
IMHO First potential human enemies to artificial intelligence creators – religious people. And “to disguise yourself a little” is a really “smart” idea."


I'm afraid that you are setting yourself up for a big surprise. Not all religious people are alike. Not all of us park our brains in the closet. I am not bluffing about any of this although I admit it helps me stay under the radar.

"General question: Is there a place for a natural selection theory within your religion beliefs?"
I believe in adaptation and intelligent design but I am not a "young earth" creationist or anything like that. I believe that life on earth was created by an advanced civilization. They are conducting an experiment of sorts. And no, I don't believe that machines will be conscious or that we'll be able to upload our minds onto computers. But there is no doubt in my mind that most of us will live to see extremely intelligent machines in our lifetimes.

Alexander B. said...

Ok. I asking this, because that moment is important.
You wrote :”I believe in adoptation”.
What do you think about a differences between human – mammal - reptilian brai?. For me, as a “Darwinist”, it seems that there should be more common parts, than differences.

Louis Savain said...

Hi Alexander,

Genetically, humans are very similar to other species, which is what we would expect from intelligent design. Even humans know to reuse proven solutions. In fact, software development is based on this principle. Complex classes are derived from less complex ones. It naturally forms a class hierarchy. Therefore, it is a sure bet that the genome of any animal will be found to be organized hierarchically with the more primitive genes at the bottom.

Louis Savain said...

I almost forgot. Here's what I think about the brain of humans and animals. I don't believe animals are conscious. Other than that, most mammal and even bird brains are similar to ours.

Alexander B. said...

Thank you. But your answers bring me to a much more questions on to your opinion. I would not ask these kind questions to the most of the others religious people. But you are really a special kind.

So, human and animals brains are very similar, but they have one general difference? How is that?
Is there some kind of special “holy” part involved, that we have, and animals don’t? Or it is just some major difference in design?
If it is a “special part”, then… Is it even possible to understand for us (for the human mind) what is conscious? Will humans discover that “special part” someday? Discover – I mean by the scientific methods, not just by made yourself believe in that.

Louis Savain said...

Every human has a spirit, of course. I am a firm believer in a Yin-Yang reality. It take two things to have consciousness: a knower and a known. Everything you see in front of you is a creation of your spirit/knower based on neuronal firings. The colorful 3-D vista that you see in front of you exists neither out there nor in your brain. It's just a bunch of neurons firing. In fact, there is no "out there" (it can be easily proven that distance is an illusion). There are no colors in your brain or in nature and beauty is not a physical property of matter and yet we recognize and see all of these things. The proof of the spirit is right in front of us, all of us.

Intelligence is at the service of motivation. It does not create motivation. It obeys it. You probably noticed that animals seem to have a certain fixed number of motivations that are closely related to survival. They have no sense of beauty that we can see and no desire to learn new things for their own sake. Humans, by contrast, find interest in things that have absolutely nothing to do with survival. Music and the arts come to mind. There is much more to human motivation than what is in our innate programming. Where is it coming from?

It goes without saying that there must be some kind of interaction between spirit and certain neurons in the brain. The problem is that a neuron in the visual cortex that is sensitive to "blue" is identical to a "red" or "green" neuron. My current take is that colors are the spirit's way of differentiating between identical but different between things. I predict that, with better tools, we will be able to pinpoint the difference between neurons that evoke conscious sensations when they are activated and those that don't.

Alexander B. said...

Are you cheating or not? I still did not decided and maybe never will. If you are, then you are a great comic. If you not, then it is surprising that a religious persons mind have such progress in AI field.
Of course the question of art is one of the most important.
IMHO cornerstone of answers will be “interest” mechanism. And “interest” mechanism is a cornerstone of attention mechanism. And there is not much difference in these principles between human and animals. There is some, but not much. Do you know that apes can speak with the finger-language? With a thousands of words in vocabulary, and with their own words created for the new purposes. And you deny them having some similar kind of consciousness?

Alexander B. said...

I did not understand about colors at all.
Many animals, even insects, have color vision. Some of them detect some colors even better than humans. For example some butterflies have great green color detection, much better than we have. They use it to find a best place to leave their caterpillars. Some special spirits in their neurons???

Louis Savain said...

I see no problem with apes and other animals having language abilities. In fact, I would expect it. But it has nothing to do with consciousness. Apes are just very intelligent. Our machines are also beginning to show a rudiment of language understanding. Does that mean they are conscious? Our future robots will probably gain a better grasp of our natural languages than most humans. I see no need for consciousness for this. Intelligence is strictly about cause-effect mechanisms.

Re: colors. We can make sensors that are sensitive to certain wavelengths. Does that mean they see colors? Of course not. There is a big difference between seeing colors and detecting frequencies. The colors that we see do not exist in the physical universe. They are in a parallel realm.

Louis Savain said...

Alexander,

Earlier you wrote:

IMHO cornerstone of answers will be “interest” mechanism. And “interest” mechanism is a cornerstone of attention mechanism. And there is not much difference in these principles between human and animals.

I agree. How and when an intelligence focuses on an object or thought obviously requires a mechanism. It's an internal motor system. I also think it is part of the motivation system of the intelligence. But in conscious humans, there is something else that is ultimately in charge of deciding what to pay attention to. It is the spirit, the true self. In future intelligent machines and in animals, this is driven entirely by a built-in, reward-punishment system.