1 / 96

Artificial Life

Artificial Life. Miriam Ruiz. Contents. Introduction Emergent Patterns Cellular Automata Agent-based modelling Distributed Intelligence Artificial Evolution Artificial Chemistry Examples Bibliography. What is Life?. There is no generally accepted definition of life .

Download Presentation

Artificial Life

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Artificial Life Miriam Ruiz

  2. Contents • Introduction • Emergent Patterns • Cellular Automata • Agent-based modelling • Distributed Intelligence • Artificial Evolution • Artificial Chemistry • Examples • Bibliography

  3. What is Life? • There is no generally accepted definition of life. • In general, it can be said that the condition that distinguishes living organisms from inorganic objects or dead organisms growth through metabolism, a means of reproduction, and internal regulation in response to the environment. • Even though the ability to reproduce is considered essential to life, this might be more true for species than for individual organisms. Some animals are incapable of reproducing, e.g. mules, soldier ants/bees or simply infertile organisms. Does this mean they are not alive? INTRODUCTION > What is Life

  4. What is Artificial Life? • The study of man-made systems that exhibit behaviors characteristic of natural livingsystems . • It came into being at the end of the ’80s when Christopher G. Langton organized the first workshop on that subject in Los Alamos National Laboratory in 1987, with the title: "International Conference on the Synthesis and Simulation of Living Systems". INTRODUCTION > What is Artificial Life

  5. What is Artificial Life? Artificial life researchers have often been divided into two main groups: • The strong alife position states that life is a process which can be abstracted away from any particular medium. • The weak alife position denies the possibility of generating a "living process" outside of a carbon-based chemical solution. Its researchers try instead to mimic life processes to understand the appearance of individual phenomena. INTRODUCTION > What is Artificial Life

  6. What is Artificial Life? • The goal of Artificial Life is not only to provide biological models but also to investigate general principles of Life. • These principles can be investigated in their own right, without necessarily having to have a direct natural equivalent. INTRODUCTION > What is Artificial Life

  7. The Basis of Artificial Life • Artificial Life tries to transcend the limitation to Earth bound life, based beyond the carbon-chain, on the assumption that life is a property of the organization of matter, rather than a property of the matter itself. INTRODUCTION > The Basis of Artificial Life

  8. The Basis of Artificial Life • Synthetic Approach: Synthesis ofcomplex systems from many simple interacting entities. • If we captured the essential spirit of ant behavior in the rules for virtual ants, the virtual ants in the simulated ant colony should behave as real ants in a real ant colony. INTRODUCTION > The Basis of Artificial Life

  9. The Basis of Artificial Life • Self-Organization: Spontaneous formation of complex patterns or complex behavior emerging from the interaction of simple lower-level elements/organisms. • Emergence: Property of a system as a whole not contained in any of its parts. Such emergent behavior results from the interaction of the elements of such system, which act following local, low-level rules. INTRODUCTION > The Basis of Artificial Life

  10. The Basis of Artificial Life • Levels of Organization: Life, as we know it on Earth, is organized into at least four levels of structure: • Molecular level. • Cellular level. • Organism level. • Population-ecosystem level. INTRODUCTION > The Basis of Artificial Life

  11. The Basis of Artificial Life • We have to distinguish between the perspective of an observer looking at an creature and the perspective of the creature itself. • In particular, descriptions of behavior from an observer's perspective must not be taken as the internal mechanisms underlying the described behavior of the creature. • The observed behavior of a creature is always the result of a system-environment interaction. It cannot be explained on the basis of internal mechanisms only. • Seemingly complex behavior does not necessarily require complex internal mechanisms. Seemingly simple behavior is not necessarily the results of simple internal mechanisms. INTRODUCTION > The Basis of Artificial Life

  12. Linear vs. Non-Linear Models • Linear models are unable to describe many natural phenomena. • In a linear model, the whole is the sum of its parts, and small changes in model parameters have little effect on the behavior of the model. • Many phenomena such as weather, growth of plants, traffic jams, flocking of birds, stock market crashes, development of multi-cellular organisms, pattern formation in nature (for example on sea shells and butterflies), evolution, intelligence, and so forth resisted any linearization; that is, no satisfying linear model was ever found. INTRODUCTION > Linear Models

  13. Linear vs. Non-linear Models • Non-linear models can exhibit a number of features not known from linear ones: • Chaos: Small changes in parameters or initial conditions can lead to qualitatively different outcomes. • Emergent phenomena: Occurrence of higher level features that weren’t explicitly modelled. • As a main disadvantage, non-linear models typically cannot be solved analytically, in contrast with Linear Models. Nonlinear modeling became manageable only when fast computers were available . • Models used in Artificial Life are always non-linear. INTRODUCTION > Non-Linear Models

  14. Contents • Introduction • Emergent Patterns • Cellular Automata • Agent-based modelling • Distributed Intelligence • Artificial Evolution • Artificial Chemistry • Examples • Bibliography

  15. Lindenmeyer Systems • Lindenmayer Systems or L-systems are a mathematical formalism proposed in 1968 by biologist Aristid Lindenmayer as a basis for an axiomatic theory on biological development. • The basic idea underlaying L-Systems is rewriting: Components of a single object are replaced using predefined rewriting rules. • Its main application field is realistic plants modelling and fractals. • They’re based in symbolic rules that define the graphic structure generation, starting from a sequence of characters. • Only as small amount of information is needed to represent very complex models. EMERGENT PATTERNS > L-Systems

  16. Lindenmeyer Systems EMERGENT PATTERNS > L-Systems

  17. Lindenmeyer Systems EMERGENT PATTERNS > L-Systems • Even though Lindenmeyer Systems do not directly generate images but long sequences of symbols, they can be interpreted in such a way that it is possible to visualize them as Turtle Graphics (Turtle Graphics were created by Seymour Papert for the LOGO language).

  18. Lindenmeyer Systems EMERGENT PATTERNS > L-Systems

  19. Diffusion Limited Aggregation (DLA) • "Diffusion limited aggregation, a kinetic critical phenomena“, Physical Review Letters, num. 47, published in 1981. • It reproduces the growth of vegetal entities like mosses, seaweed or lichen, and chemical processes such as electrolysis or the crystallization of certain products. • A number of moving particles are freed inside an enclosure where we have already one or more particles fixed. • Free particles keep moving in a Brownian motion until they reach a fixed particle nearby. In that case they fix themselves too. EMERGENT PATTERNS > DLA

  20. Diffusion Limited Aggregation (DLA) EMERGENT PATTERNS > DLA

  21. Diffusion Limited Aggregation (DLA) EMERGENT PATTERNS > DLA

  22. Diffusion Limited Aggregation (DLA) EMERGENT PATTERNS > DLA

  23. Diffusion Limited Aggregation (DLA) EMERGENT PATTERNS > DLA

  24. Contents • Introduction • Emergent Patterns • Cellular Automata • Agent-based modelling • Distributed Intelligence • Artificial Evolution • Artificial Chemistry • Examples • Bibliography

  25. Cellular Automata • Discrete model studied in computability theory and mathematics. • It consists of an infinite, regular grid of cells, each in one of a finite number of states. • The grid can be in any finite number of dimensions. • Time is also discrete, and the state of a cell at time t is a function of the state of a finite number of cells called the neighborhood at time t-1. • The neighbourhood is a selection of cells relative to some specified, and does not change. • Every cell has the same rule for updating, based on the values in this neighbourhood. • Each time the rules are applied to the whole grid a new generation is produced. CELLULAR AUTOMATA > Introduction

  26. Wolfram’s Cellular Automata CELLULAR AUTOMATA > Wolfram CAs • Studied by Stephen Wolfram at the beginning of the ’80s. • Unidimensional cellular automata with a neighbourhood of 1 cell around the one we’re studying. • There are 256 elemental Wolfram CAm each of them with an associated “Wolfram Number”.

  27. Wolfram’s Cellular Automata CELLULAR AUTOMATA > Wolfram CAs

  28. Wolfram’s Cellular Automata CELLULAR AUTOMATA > Wolfram CAs

  29. Wolfram’s four Classes of CA • Class I (Empty): Tends to spatially homogeneous state (all cells are in the same state). Patterns disappear with time. Small changes in the initial conditions cause no change in final state. • Class II (Stable or Periodic): Yields a sequence of simple stable or periodic structures (endless cycle of same states). Point attractor or periodic attractor. Small changes in the initial conditions cause changes only in a region of finite size. • Class III (Chaotic): Exhibits chaotic aperiodic behavior. Pattern grows indefinitely at a fixed rate. Small changes in the initial conditions cause changes over a region of ever-increasing size. • Class IV (Complex): Yields complicated localized structures, some propagating. Pattern grows and contracts with time. Small changes in the initial conditions cause irregular changes. CELLULAR AUTOMATA > Wolfram CAs

  30. Class IV CA Examples CELLULAR AUTOMATA > Wolfram CAs

  31. 1-D CA Example: Seashells CELLULAR AUTOMATA > Wolfram CAs

  32. Conway’s Game of Life • Invented by english mathematician John Conway and published by Martin Gardner in Scientific American in 1970. • Bidimensional board, in each cell can be one or none live cells (binary). • The neighbourhood is the 8 surrounding cells. • Very simple rule set: • Survival: A cell survives if there are 2 or 3 live cells in its neighbourhood. • Death: A cell surrounded by other 4 or more dies of overpopulation. If it is surrounded by one or none, dies of isolation. • Birth: An empty place surrounded by exactly three cells gives place to a new cell’s birth. • The result is a Turing-Complete system. CELLULAR AUTOMATA > Conway’s Game of Life

  33. Conway’s Game of Life CELLULAR AUTOMATA > Conway’s Game of Life

  34. Conway’s Game of Life CELLULAR AUTOMATA > Conway’s Game of Life

  35. Contents • Introduction • Emergent Patterns • Cellular Automata • Agent-based modelling • Distributed Intelligence • Artificial Evolution • Artificial Chemistry • Examples • Bibliography

  36. Agent-based Modelling • Computational model based in the analysis of specific individuals situated in an environment, for the study of complex systems. • The model was conceptually developed at the end of the ’40s, and had to wait for the arrival of computers to be able to develop totally. • The idea is to build the agents, or computational devices, and simulate them in parallel to be able to model the real phenomena that is being analysed. • The resulting process is the emergency from lower levels of the social system (micro) towards the upper levels (macro). AGENTS > Introduction

  37. Agent-based Modelling • Simulations based in agents have two essential components: • Agents • Environment • The environment has a certain autonomy from the actions of the agents, although it can be modified by their behaviour. • The interaction between the agents is simulated, as well as the interaction between the agents and their surrounding environment. AGENTS > Introduction

  38. Artificial Societies: Chimps • Charlotte Hemelrijk has investigated (1998) the emergence of structure in societies of primates in the real world and in simulation. • Her creatures were able to move and to see each other. If creatures perceived someone nearby, they engaged in dominance interactions. • The effects of losing (and winning) are self-reinforcing: after losing a fight the chance to loose the next fight is larger (even if the opponent is weak). The winner effect is the converse. • If they were not engaged in dominance interactions, they followed rules of moving and turning, that kept them aggregated (because real primates are group-living). • It is unnecesary to consider the representation of a hierarchical structure in the individual minds of the chimps, because it appears spontaneously as an emergent structure of the group. AGENTS > Chimps

  39. Artificial Societies: Chimps AGENTS > Chimps

  40. Artificial Societies: Chimps • Interactions among these artificial chimps are just triggered by the proximity of others not by record keeping or other strategic considerations. • A dominance hierarchy arose, and a social-spatial structure, with dominants in the center and subordinates at the periphery, similar to what has been described for several primate species. • For an external observer, support in fights appeared to be repaid, despite the absence of a motivation to support or keep records of them. • This was a consequence of the occurrence of a series of cooperation that consisted of two creatures alternatively supporting each other to chase away a third. • These originated because by fleeing from the attack range of one opponent the victim ended up in the attack range of the other opponent. This typically ended when the spatial structure had changed such that one of both cooperators attacked the other. AGENTS > Chimps

  41. Artificial Societies: Chimps AGENTS > Chimps

  42. Contents • Introduction • Emergent Patterns • Cellular Automata • Agent-based modelling • Distributed Intelligence • Artificial Evolution • Artificial Chemistry • Examples • Bibliography

  43. Distributed Intelligence • Complex behaviour patterns of a group, in which there is no central command. • It arises from “emergent behaviour”. • It appears in a group as a whole, but is no explicitly programmed in none of the individual members of the group. • Simple behaviour rules in the individual members of the group can cause a complex behaviour pattern of the group as a whole. • The group is able to solve complex problems a partir only local information. • Examples: Social insects, immunological system, neural net processing. DISTRIBUTED INTELLIGENCE > Introduction

  44. Didabots • Experiment carried on in 1996, studying the collective behaviour of simple robots, called Didabots. • The main idea is to verify that apparently complex behaviour patterns can be a consequence of very simple rules that guide the interactions between the entities and the environment. • This idea has been successfully applied for example to the study of social insects. DISTRIBUTED INTELLIGENCE > Didabots

  45. Didabots • Infrared sensors can be used to detect proximity up to about 5 cm. • Programmed exclusively for avoiding obstacles. • Sensorial stimulation of the left sensor makes the bot turn a bit to the right, and viceversa. DISTRIBUTED INTELLIGENCE > Didabots

  46. Didabots DISTRIBUTED INTELLIGENCE > Didabots

  47. Didabots • Initially the cubes are randomly distributed. • Over time, a number of clusters start to form. In the end, there are only two clusters and a number of cubes along the walls of the arena. • These experiments were performed many times and the result is very consistent. • Apparently Didabots are cleaning the arena, grouping blocks into clusters, from an external observer point of view. • The robots were only programmed to avoid obstacles. • This happens because when there is a cube right in front of the Didabot, it is not able to detect it, and thew Didabot pushes the cube until it collides with another cube. The cube being pushed is slightly moved and it enters the perception space of one of the sensors. The Didabot turns a bit then and leaves the cube. DISTRIBUTED INTELLIGENCE > Didabots

  48. Social Insects • The main quality for the so-called social insects, ants or bees, is to form part of a self-organised group, whose key aspect is “simplicity”. • These insects solve their complex problems through the sum of simple interactions of every individual insect. DISTRIBUTED INTELLIGENCE > Social Insects

  49. Bees • The distribution of brood and nourishment in the comb of honey bees is not random, but forms a regular pattern . • The central brooding region is close to a region containing pollen and one containing nectar (providing protein and carbohydrates for the brood). • Due to the intake and outtake of pollen and nectar, the pattern is changing all the time on a local scale, but it stays stable if observed from a more global scale. DISTRIBUTED INTELLIGENCE > Social Insects

  50. Bees • This is not the result of an individual bee being aware of the global pattern of brood- and food-distribution in the comb, but of three simple local rules, which each individual bee follows: • Deposit brood in cells next to cells already containing brood. • Deposit nectar and pollen in discretionary cells but empty the cells closest to the brood first. • Extract more pollen than nectar. DISTRIBUTED INTELLIGENCE > Social Insects

More Related