Models of artificial life. Part 2
At the request of the habravchan continuation of part 1
It is not a secret for anybody that for the entire animal world, including man, there is an inherent focus of behavior, i.e. the desire to achieve certain goals. For animals, it is most often survival and reproduction. In their work, Burtsev, Gusarev and Redko try to use the computer model to answer the question: “How could purposeful behavior in principle arise in the process of the evolution of life on our planet?”.
In this project, the authors use such a concept as motivation . After all, the motivation of any living creature that falls into one or another situation stimulates the “right” decision making. Using motivation, the model explores a possible mechanism for the emergence of purposeful behavior in the process of evolution.
All actions in the model take place in a one-dimensional cellular environment. Time is discrete - i.e. one action is performed at a time. With cells, with a certain probability, grass grows. There is a population of agents that have a need for energy, due to nutrition and the need for reproduction. The energy of an agent is spent when performing the following actions, and a different amount of energy is spent on performing different actions:
Each agent need is quantified by motivation. For example, if an agent sees another agent nearby and his energy resource is sufficient for reproduction, he marks himself as ready for reproduction, if the second agent does the same, a crossing occurs. As a result, a new agent appears who takes parts of the energy resource from the parents. Each agent has its own neural network, which has special inputs from motivations. At the expense of the neural network, the behavior of the agent is controlled and the evolution of agents takes place - the genome (set of weights of the neural network) of the descendant is formed, as a result of crossing, on the basis of the parents' genome using recombination and mutation. When reducing the resource to zero - the agent dies. Two variants were modeled for analysis: agents with motivation and agents without motivations. There were also given different parameters P - the probability of an accidental appearance of grass in each time. The final results showed that the population of agents with motivations is much better adapted to the environment, and with an average amount of food (P = 1/200), the population of agents with motivation “finds” a rather effective survival strategy, and the population without motivations dies out completely.
The results are logical and very understandable, because for an agent without motivation, all that remains is to eat it at the sight of food, to cross with it at the sight of a neighbor, to stand in the absence of everything and do nothing (rest), which leads to imminent death, with a small or medium amount of food. When an agent has a motivation, in the process of evolution, the agent begins to act approximately according to this scheme: there is little resource - to search for food or rest, a lot of resource - to perform any actions. Due to this scheme, the population survives much more efficiently than in the case without motivations.
Many, watching the behavior of ants, pay attention to the work of ants. Harvesting food, they take it to an anthill, where it is processed and eaten by all members of the colony. Many species of ants have a high degree of coordination and cooperation between “miners” (usually by means of pheromone information transfer).
"Ant farm" imitates the evolution of complex behavior in artificial organisms. It examines organisms that live and reproduce in a relatively complex environment, with a multitude of feelings (internal and external), and a large set of possible actions at each point in time. Moreover, each ant has a memory, and, consequently, its behavior depends on its history. During the life of each ant is born, has time to make thousands of different decisions and actions, leaves offspring (it depends on the behavior of the organism throughout its life), and eventually dies.
By the way, the AntFarm model was developed in the CM ++ language based on the supercomputer Connection Machine 2 , which specializes in developments in the field of Artificial Intelligence.
')
UPD Article , in English. It describes more or less in detail the model of Collins and Jefferson - AntFarm.
Grasshopper
Authors: M. S. Burtsev, R. V. Gusarev, V. G. Redko, 2002
It is not a secret for anybody that for the entire animal world, including man, there is an inherent focus of behavior, i.e. the desire to achieve certain goals. For animals, it is most often survival and reproduction. In their work, Burtsev, Gusarev and Redko try to use the computer model to answer the question: “How could purposeful behavior in principle arise in the process of the evolution of life on our planet?”.
In this project, the authors use such a concept as motivation . After all, the motivation of any living creature that falls into one or another situation stimulates the “right” decision making. Using motivation, the model explores a possible mechanism for the emergence of purposeful behavior in the process of evolution.
All actions in the model take place in a one-dimensional cellular environment. Time is discrete - i.e. one action is performed at a time. With cells, with a certain probability, grass grows. There is a population of agents that have a need for energy, due to nutrition and the need for reproduction. The energy of an agent is spent when performing the following actions, and a different amount of energy is spent on performing different actions:
- To be at rest (rest) - the least amount of energy;
- There is (to replenish your energy resource) - twice the amount of energy;
- Move, i.e. move one cell to the right or left - another two times more energy costs than on food;
- Jumping through several cells in a random direction - 5 times more than when moving;
- To interbreed - also, 5 times more energy costs than when moving.
Each agent need is quantified by motivation. For example, if an agent sees another agent nearby and his energy resource is sufficient for reproduction, he marks himself as ready for reproduction, if the second agent does the same, a crossing occurs. As a result, a new agent appears who takes parts of the energy resource from the parents. Each agent has its own neural network, which has special inputs from motivations. At the expense of the neural network, the behavior of the agent is controlled and the evolution of agents takes place - the genome (set of weights of the neural network) of the descendant is formed, as a result of crossing, on the basis of the parents' genome using recombination and mutation. When reducing the resource to zero - the agent dies. Two variants were modeled for analysis: agents with motivation and agents without motivations. There were also given different parameters P - the probability of an accidental appearance of grass in each time. The final results showed that the population of agents with motivations is much better adapted to the environment, and with an average amount of food (P = 1/200), the population of agents with motivation “finds” a rather effective survival strategy, and the population without motivations dies out completely.
The results are logical and very understandable, because for an agent without motivation, all that remains is to eat it at the sight of food, to cross with it at the sight of a neighbor, to stand in the absence of everything and do nothing (rest), which leads to imminent death, with a small or medium amount of food. When an agent has a motivation, in the process of evolution, the agent begins to act approximately according to this scheme: there is little resource - to search for food or rest, a lot of resource - to perform any actions. Due to this scheme, the population survives much more efficiently than in the case without motivations.
AntFarm
Authors: Robert J. Collins, David R. Jefferson, 1991
Many, watching the behavior of ants, pay attention to the work of ants. Harvesting food, they take it to an anthill, where it is processed and eaten by all members of the colony. Many species of ants have a high degree of coordination and cooperation between “miners” (usually by means of pheromone information transfer).
"Ant farm" imitates the evolution of complex behavior in artificial organisms. It examines organisms that live and reproduce in a relatively complex environment, with a multitude of feelings (internal and external), and a large set of possible actions at each point in time. Moreover, each ant has a memory, and, consequently, its behavior depends on its history. During the life of each ant is born, has time to make thousands of different decisions and actions, leaves offspring (it depends on the behavior of the organism throughout its life), and eventually dies.
By the way, the AntFarm model was developed in the CM ++ language based on the supercomputer Connection Machine 2 , which specializes in developments in the field of Artificial Intelligence.
')
UPD Article , in English. It describes more or less in detail the model of Collins and Jefferson - AntFarm.
- Part 1 (PolyWorld, Interaction between learning and Evolution)
- Part 2 (Grasshopper, AntFarm)
- Part 3 - coming soon ...
Source: https://habr.com/ru/post/55224/
All Articles