May 18, 2011 at 18:28Memory complication in neural networks
While reading the article “21st century: what is life from the point of view of physics” I came across a description of memory in both living and non-living matter. By memory in living matter, everyone understands what is meant, but by memory in non-living, all means of storing information in your computer are meant and many other ways. So, from the point of view of the author of the article, G. Ivanitsky, a distinctive feature of life is the use of memory for forecasting. Everything would be fine, but only we have already created many forecasting programs and machines that are guided by such programs. I don’t want to raise at least a few philosophical questions about where is the boundary between the living and the non-living, are the robots alive then, and we can just complex composite automata, etc. And just bring a line of reasoning to a problem that interests me more,
Darwin suggested that the driving force behind evolution is heredity, selection, and variability. However, selection cannot be a driving force; it comes from what already exists, reducing diversity. Selection is a reduction operation. Instead of the term “variability,” the term “self-complication” should be used. But how is the self-complication mechanism implemented?
As for any process, energy and a guide for the use of this energy are necessary for complication. The author does not give a direct answer as to what should happen here. He draws the closest possible analogy in his opinion to this process. As from a random symmetric chaotic process, a directed process and energy storage arise. Why is a symmetrical chaotic process, because the occurrence of life is considered at the molecular level in a liquid characterized by thermodynamic equilibrium. And for simplicity and clarity of explanation, the symmetric chaotic process is reduced to a probabilistic binary game.
A probabilistic binary game is a simple coin toss. Such a game is symmetrical; for an infinite number of throws, the average probabilities of winning and losing are equal. The question is, can a player break the symmetry in any way?
The conditions of the game are quite strict, every time a coin is flipped, the game starts again, all previous game results are forgotten. The probability of each new game does not depend on the previous one. This means that there is no strategy to win, and the optimal strategy, contrasting the chance of one side of the coin falling out with a completely random side guessing strategy, guarantees a draw.
Nevertheless, if a player has at least one round of memory and the ability to change the bet, then you can choose a strategy for changing the bet that brings asymmetry to the player. When considering the results of games in time, they can be represented as a sequence of alternating clusters of wins and losses of various lengths. The strategy is to increase the rate from one by arithmetic progression as soon as you win. In total, after applying this strategy, the winning clusters will bring more than the losing clusters will take.
Such a strategy is the simplest example of a control system with a variable feedback coefficient. In our case, the player came up with a strategy, it is unknown how this happened in nature. But it is obvious that the central idea of all living objects is a survival strategy that requires a control system. In living organisms, such systems are implemented both at the molecular level in unicellular and at the cellular level - on neurons in the nervous system of higher life forms.
The control system described above has a memory used to store the result of the last coin flip game. With each new game, the memory cell updates its value. Such a memory is based on a simple structure and is easily implemented on neural networks, but nevertheless it holds exactly one clock cycle. And perhaps this was the first step of nature in the evolution of life management systems.
Now the most complex product of evolution is the human brain, which consists of many neurons. Neurons and connections between neurons are very, very complex structured. The brain map describing what are the structural parts of the brain and what they are responsible for is still being refined and improved. And we are far from even drawing up a “connect” of the human brain - a complete description of the structure of connections in the nervous system of the body.
But one of the important properties of the human brain is “permanent memory”, which allows you to learn and implement the above and similar strategies without owning them from birth. It would be logical to assume that the structure of our brain allows us to do such tricks. And if we maximally discard all the complicating properties and set ourselves the task of creating a control system model, the structure of which would allow us to keep the simplest strategy in our memory and at the same time be guided by it. (I am struck by a strict analogy with the expression "the system is aware of what it is doing.") Can we solve such a problem on neurons? As such, a solution already exists, but in the form of expert systems, programs that derive new theorems and the like, but not on neurons. And I have not heard that someone would seriously engage in the construction of such structures on neural networks.connectom . But in the worm there is no structure of “permanent” memory; only higher forms of life have this property. Or, from my point of view, a poorly conscious simulation of the simultaneous operation of neurons in amounts a little more than in an ordinary cat .
I have an article about how I modeled on a neural network a simple amoeba control system that goes towards the goal. Now I want to build a control structure on neurons that would meet the following requirements:
From my previous experience, I can say that the task will probably be simplified even further to the most primitive level while maintaining the central idea of realizing “permanent” memory. Well, the result will be in the form of another way how it does not work.
Materials:
Article “ 21st century: what is life from the point of view of physics ” G.R. Ivanitsky in the journal Uspekhi Fizicheskikh Nauk.
Video program " Popular Science " with a discussion of the article.
Darwin suggested that the driving force behind evolution is heredity, selection, and variability. However, selection cannot be a driving force; it comes from what already exists, reducing diversity. Selection is a reduction operation. Instead of the term “variability,” the term “self-complication” should be used. But how is the self-complication mechanism implemented?
As for any process, energy and a guide for the use of this energy are necessary for complication. The author does not give a direct answer as to what should happen here. He draws the closest possible analogy in his opinion to this process. As from a random symmetric chaotic process, a directed process and energy storage arise. Why is a symmetrical chaotic process, because the occurrence of life is considered at the molecular level in a liquid characterized by thermodynamic equilibrium. And for simplicity and clarity of explanation, the symmetric chaotic process is reduced to a probabilistic binary game.
A probabilistic binary game is a simple coin toss. Such a game is symmetrical; for an infinite number of throws, the average probabilities of winning and losing are equal. The question is, can a player break the symmetry in any way?
The conditions of the game are quite strict, every time a coin is flipped, the game starts again, all previous game results are forgotten. The probability of each new game does not depend on the previous one. This means that there is no strategy to win, and the optimal strategy, contrasting the chance of one side of the coin falling out with a completely random side guessing strategy, guarantees a draw.
Nevertheless, if a player has at least one round of memory and the ability to change the bet, then you can choose a strategy for changing the bet that brings asymmetry to the player. When considering the results of games in time, they can be represented as a sequence of alternating clusters of wins and losses of various lengths. The strategy is to increase the rate from one by arithmetic progression as soon as you win. In total, after applying this strategy, the winning clusters will bring more than the losing clusters will take.
Such a strategy is the simplest example of a control system with a variable feedback coefficient. In our case, the player came up with a strategy, it is unknown how this happened in nature. But it is obvious that the central idea of all living objects is a survival strategy that requires a control system. In living organisms, such systems are implemented both at the molecular level in unicellular and at the cellular level - on neurons in the nervous system of higher life forms.
The control system described above has a memory used to store the result of the last coin flip game. With each new game, the memory cell updates its value. Such a memory is based on a simple structure and is easily implemented on neural networks, but nevertheless it holds exactly one clock cycle. And perhaps this was the first step of nature in the evolution of life management systems.
Now the most complex product of evolution is the human brain, which consists of many neurons. Neurons and connections between neurons are very, very complex structured. The brain map describing what are the structural parts of the brain and what they are responsible for is still being refined and improved. And we are far from even drawing up a “connect” of the human brain - a complete description of the structure of connections in the nervous system of the body.
But one of the important properties of the human brain is “permanent memory”, which allows you to learn and implement the above and similar strategies without owning them from birth. It would be logical to assume that the structure of our brain allows us to do such tricks. And if we maximally discard all the complicating properties and set ourselves the task of creating a control system model, the structure of which would allow us to keep the simplest strategy in our memory and at the same time be guided by it. (I am struck by a strict analogy with the expression "the system is aware of what it is doing.") Can we solve such a problem on neurons? As such, a solution already exists, but in the form of expert systems, programs that derive new theorems and the like, but not on neurons. And I have not heard that someone would seriously engage in the construction of such structures on neural networks.connectom . But in the worm there is no structure of “permanent” memory; only higher forms of life have this property. Or, from my point of view, a poorly conscious simulation of the simultaneous operation of neurons in amounts a little more than in an ordinary cat .
I have an article about how I modeled on a neural network a simple amoeba control system that goes towards the goal. Now I want to build a control structure on neurons that would meet the following requirements:
- an amoeba, entering a room with two landmarks, randomly walking around the room, stumbles upon hidden food and randomly located relative to the landmarks.
- then the amoeba gets into another room without the above guidelines, but with any others.
- then the amoeba again enters the first room and first goes to the old place where she found food.
From my previous experience, I can say that the task will probably be simplified even further to the most primitive level while maintaining the central idea of realizing “permanent” memory. Well, the result will be in the form of another way how it does not work.
Materials:
Article “ 21st century: what is life from the point of view of physics ” G.R. Ivanitsky in the journal Uspekhi Fizicheskikh Nauk.
Video program " Popular Science " with a discussion of the article.