Collective Intelligence: from ants to neurons

1. Collective Intelligence: from ants to neurons “Dumb parts, properly connected into a swarm, yield to smart results” IFAE-Thursday Meeting, 22nd February 2007 Estel Pérez

2. What is this all about? • "An individual ant is not very bright, but ants in a colony, operating as a collective, do remarkable things. A single neuron in the human brain can respond only to what the neurons connected to it are doing, but all of them together can be Albert Einstein."By Deborah M. Gordon (Stanford University) • We are interested in systems where simple units together behave in complicated ways.

3. Outline • Introduction • Complexity • Emergence • Examples • Swarm Intelligence: learning from Nature • Ants • Natural Ants: How do they do it? • Ant Colony Optimization • Applications: TSP • Birds & Fish • Modeling Bird Flocking • Particle Swarm Optimization • Applications • Conclusions

4. Complexity • Artificial Intelligence • Neural Networks • Chaos • Butterfly Effect • Attractors • Fractals • Self-Organization • Non-linear systems • Emergence • Collective Intelligence

5. Complexity • Studies systems with many strongly-coupled degrees of freedom. • Many natural, artificial and abstract objects or networks can be considered to be complex systems. • The study of complexity is highly interdisciplinary. • Examples of complex systems include ant-hills, human economies, climate, nervous systems, cells and living things, including human beings, as well as modern telecommunication infrastructures.

6. Complexity • All complex systems have behavioral and structural features in common: • Relationships are non-linear • Relationships contain feedback loops • Complex systems have hysteretic behavior: they change over time, and prior states may have an influence on present states. • Complex systems may be nested: The components of a complex system may themselves be complex systems. (cell-organism-colony-ecosystem-Gaia) • May produce emergent phenomena.

7. Emergence • "The whole is more than the sum of its parts“ (Aristotle) • Definition: "the arising of novel and coherent structures, patterns and properties during the process of self-organization in complex systems." By Goldstein • “Superficial complexity that arises from a deep simplicity” by Murray Gell-Mann (Nobel Prize for the quark model) • Bottom-up behavior: Simple agents following simple rules generate complex structures/behaviors. • Agents don’t follow orders from a leader. A termite "cathedral" mound produced by a termite colony: a classic example of emergence in nature.

8. Examples • Biology: The cellular slime molds are unicellular organisms that usually take the form of individual amoebae, but under stress they aggregate to form a multicellular assembly. Model: • Slime mold gives off a substance called pheromone all the time. • Local Rules: • Move in the direction that has the highest pheromone concentration. • If no pheromone, move randomly. • All the while, each slime mold cell is giving off a pheromone which evaporates at each time step. • Parameters: number of cells, rate of pheromone evaporation

9. Examples • Economics: Stock market precisely regulates the relative prices of companies across the world, yet it has no leader. • The World Wide Web is a decentralized system exhibiting emergent properties. The number of links pointing to each page follows a power law. • Mathematics: a Moebius strip has emergent properties: it can be constructed from a set of two-sided, four edged, squared surfaces. Only the complete set of squares is one-sided and one-edged! • Could Human Conscience be explained as an emerging behavior from the interaction of individual neurons ?

10. Swarm Intelligence • Based on the study of emergent collective intelligence of groups of simple agents Bird Flock Animal Herd Ant Colony Fish School

11. Learning from Nature • Nature has inspired researchers in many different ways. • Airplanes have been designed based on the structures of birds' wings. • Robots have been designed in order to imitate the movements of insects. • Resistant materials have been synthesized based on spider webs. • After millions of years of evolution all these species developed solutions for a wide range of problems. Some ideas can be developed by taking advantage of the examples that Nature offers.

12. Learning from Nature • Some social systems in Nature can present an intelligent collective behavior although they are composed by simple individuals. • The intelligent solutions to problems naturally emerge from the self-organization and communication of these individuals. • These systems provide important techniques that can be used in the development of distributed artificial intelligent systems.

13. Swarm Intelligence • Swarm Intelligence is an artificial intelligence technique based on the study of collective behavior in self-organized systems. • Swarm Intelligence systems are typically made up of a population of simple agents interacting locally with one another and with their environment. This interaction often lead to the emergence of global behavior. • The main bio-inspired algorithms that have been developed are: • Ant Colony Optimisation (ACO) • Particle Swarm Optimisation (PSO)

14. Natural Ants • Individual ants are simple insects with limited memory and capable of performing simple actions. • However, an ant colony expresses a complex collective behavior providing intelligent solutions to problems such as: • carrying large items • forming bridges • finding the shortest routes from the nest to a food source, prioritizing food sources based on their distance and ease of access.

15. Natural Ants • Moreover, in a colony each ant has its prescribed task, but the ants can switch tasks if the collective needs it. • Outside the nest, ants can have 4 different tasks: • Foraging: searching for and retrieving food • Patrolling: looking for food supply • Midden work: Sorting the colony refuse pile • Nest maintenance work: construction and clearing of chambers • An ant’s decision whether to perform a task depends on: • The Phisical State of the environment: • If part of the nest is damaged, more ants do nest maintenance work to repair it • Social Interactions with other ants

16. How Do Social Insects AchieveSelf-organization? • Communication is necessary • Two types of communication: • Direct: antennation, trophallaxis (food or liquid exchange), mandibular contact, visual contact, chemical contact, etc. • Indirect: two individuals interact indirectly when one of them modifies the environment and the other responds to the new environment at a later time. This is called stigmergy.

17. Natural ants: How do they do it? • How do they know which task to perform? • When ants meet, they touch with their antennae, that are organs of chemical perception. • An ant can perceive the colony-specific odor that all nest mates share. • In addition to this odor, ants have an odor specific to their task, because of the temperature and humidity conditions in which it works. • So that an ant can evaluate its rate of encounter with ants of a certain task. • The pattern of interaction each ant experiences influences the probability it will perform a task.

18. Natural ants: How do they do it? • How can they manage to find the shortest path? "The best possible way for ants to find anything is to have an ant everywhere all the time, because if it doesn't happen close to an ant,  they are not going to know about it. Of course there are not enough ants in the colony, so the ants have to move around in a pattern that allows them to cover space efficiently"

19. Natural ants: How do they do it? • They establish indirectcommunication system based on the deposition of pheromone over the path they follow. • An isolated ant moves at random, but when it finds a pheromone trail, there is a high probability that this ant will decide to follow the trail. • An ant foraging for food deposits pheromone over its route. When it finds a food source, it returns to the nest reinforcing its trail. • So, other ants have greater probability to start following this trail and thereby laying more pheromone on it. • This process works as a positive feedback loop system because the higher the intensity of the pheromone over a trail, the higher the probability of an ant start traveling through it.

20. Since the route B is shorter, the ants on this path will complete the travel more times and thereby lay more pheromone over it. The pheromone concentration on trailB will increase at a higher rate than on A, and soon the ants on route A will choose to follow route B Since most ants will no longer travel on route A, and since the pheromone is volatile, trail A will start evaporating Only the shortest route will remain! Natural ants: How do they do it?

21. (1) Ants finished all using the same path (each one of the 2 paths, 50% of times) (2) Ants use the short path (3) Ants get to find the shortest path Natural ants: Experiments (1) (2) (3) (1) (2)

22. Modeling Ants Colony • It is known that the ability of ants in finding the shortest route between the nest and a food source can be used to solve graph problems. • Environment: • Actions that an agent performs: • In a city, it chooses a route based on the intensity of the pheromone over the available paths • When it finds the food source, it starts the return travel on its own pheromone trail • All actions require only local information and short memory

23. Ant Colony Optimization • Optimization Technique Proposed by Marco Dorigo in the early ’90 • Each artificial ant is a probabilistic mechanism that constructs a solution to the problem, using: • Artificial pheromone deposition • Heuristic information: pheromone trails, already visited cities memory … • Differences between real and artificial ants: • Artificial ants live in a discrete world • The pheromone is updated only after a solution has been constructed. • Additional mechanisms

24. Ant Colony Optimization • ConstructAntSolutions • The exact rules for the probabilistic choice of solution components vary across different ACO variants. • UpdatePheromones • It is used to increase the pheromone values associated with good or promising solutions, and decrease those that are associated with bad ones. • Decreasing all the pheromone values through pheromone evaporation -> allows “forgetting”-> favors exploration of new areas • Increasing the pheromone levels associated with a chosen set of good solutions -> makes the algorithm converge to a solution • Supdis the set of solutions that are used for the update • ρ (0; 1] is a parameter called evaporation rate • F is a function commonly called the fitness function.

25. Ant Colony Optimization • Different ACO algorithms differ in the way they update the pheromone. • AS-update: Supd = Siter (the set of solutions that were constructed in the current iteration) -> Like in Nature • IB-update: Supd = Sib = arg max F(s) (iteration-best solution: the best solution in the current iteration) • introduces a much stronger bias towards the good solutions -> increases speed • Increases the probability of premature convergence • BS-update: Supd = Sbs(best-so-far solution: the best solution since the first algorithm iteration) • Introduces an even stronger bias • In practice, ACO algorithms that use variations of the IB-update or the BS-update rules and that additionally include mechanisms to avoid premature convergence, achieve better results than those that use the AS-update rule.

26. Applications • The ACO can be used to solve graph problems such as the Traveling Salesman Problem (TSP). • Of High computational complexity • For which the exact algorithms are inefficient • For which we don’t need the best solution but a good one.

27. If a salesman starts at point A, and if the distances between every pair of points are known, what is the shortest route which visits all points and returns to point A? Traveling Salesman Problem • Given a number of cities and the costs of traveling from any city to any other city, what is the cheapest round-trip route that visits each city exactly once and then returns to the starting city? • Trying all possible solutions means n! permutations. • Using the techniques of dynamic programming, it can be solved in time O(n22n) • The problem is of considerable practical importance. Example: printed circuit manufacturing: scheduling of a route of the drill machine to drill holes in a PCB.

28. TSP solved using ACO I shoud have an applet, but…

29. Modeling bird flocking • The synchrony of flocking behavior is thought to be a function of bird’s efforts to maintain an optimal distance between themselves and their neighbors. “Individual members can profit from the discoveries and previous experience of other members during the search for food. This advantage can become decisive , overweighting the disadvantages of competition for food” • Birds and fish adjust their physical movement to avoid predators, seek for food and mates. • Humans tend to adjust our beliefs and attitudes to conform with those of our social peers. Humans change in abstract multidimensional space, collision-free.

30. Modeling bird flocking • Definitions: • Flock is a group of objects that exhibit the general class of polarized (aligned), non-colliding, aggregate motion. • Boid is a simulated bird-like object, i.e., it exhibits this type of behavior. It can be a fish, bee, dinosaur, etc. • Rules for flocking: • Cohesion: Each boid fly towards the centroid of its local flock mates (that is, boid in its local neighborhood) • Separation : Each boid keep a certain distance away from local flock mates to avoid collisions • Alignment: Each boid align its velocity vector and keep velocity magnitude similar with that of the local flock • Note: There might be many other rules for making the flock more realistic.

31. Particle Swarm Optimisation • Proposed by Eberhart and Kennedy in the middle ’90 • Global Optimization Algorithm dealing with problems in which a best solution can be represented as a point or surface in an n-dimensional space. • Inspired in the social behavior of bird flocks and fish schools • The main application is on Numeric Optimization • Advantages: • It requires only primitive mathematical operators • Is computationally inexpensive, in terms of both • Memory requirements • Speed • The large number of members that make up the particle swarm make the technique impressively resistant to the problem of local minima.

32. Particle Swarm Optimisation • Imagine a bird’s flock in an area where there is a single food source. • A bird don’t know where the food is, but it knows its distance to the food. • The best strategy is to follow the bird that is closer to the food. • Particles save and communicate the best solution they have found.

33. Particle Swarm Optimisation • It considers a particle swarm (or cloud) that moves over the solution space, and particles are evaluated according to some fitness criterion. The movement of each particle depends on: • Its best position since the algorithm started (pBest) • The best position of the particles around it (lBest) or of the whole group (gBest) • On each iteration, the particle changes its velocity towards pBest and lBest/gBest. • So the swarm explores the solution space looking for promising zones.

34. Particle Swarm Optimisation • The pseudo code of the procedure is as follows:For each particle     Initialize particleENDDo     For each particle         Calculate fitness value        If the fitness value is better than the best fitness value (pBest) in history            set current value as the new pBest    End    Choose the particle with the best fitness value of all the particles as the gBest     For each particle         Calculate particle velocity according equation (a)        Update particle position according equation (b)    End While maximum iterations or minimum error criteria is not attained

35. Particle Swarm Optimisation • Combination of gBest and the pBest : need a compromise • lBest can be: • Social: the particles around are always the same, no matter where they are in space • Geographical: the particles around are those whose distance is the shortest • Global PSO vs. Local PSO: the global version converges quickly to a solution but it gets more easily stuck in local minima.

36. Swarm Technology:Applications • Swarm technology is particularly attractive because it is • cheap • robust • Simple • Some examples of applications: • Controlling unmanned vehicles • Possibility of using it to control nanobots within the body to kill cancer tumors • Disney's The Lion King was the first movie to make use of swarm technology (the stampede of the wildebeests scene). The Lord of the rings used it too during the battle scenes. • Grid Data Replication:

37. Conclusions • We can learn from nature and take advantage of the problems that she has already solved. • Many simple individuals interacting with each other can make a global behavior emerge. • Techniques based on natural collective behavior (Swarm Intelligence) are interesting as they are cheap, robust, and simple. • They have lots of different applications. • Swarm intelligence is an active field in Artificial Intelligence, many studies are going on.

38. That’s all! Thank you for your attention