Swarm Intelligence: A new way to think about business

1. Swarm Intelligence:A new way to think about business Professor Kesheng Wang Department of Production and Quality engineering Norwegian University of Science and Technology, Trondheim Norway knowledge management, Calpe

2. Outlines • Introduction • Foraging for solution • The task of dividing tasks • Simple rules rule • Raiding new markets • A swarm of possibilities • Conclusion knowledge management, Calpe

3. 1. Introduction • Swarm Smarts/Intelligence are based Natural examples, which include ant colonies, bird flocking, animal herding, bacterial growth, and fish schooling. • Insects that live in colonies — ants, bees, wasps, termites birds and fish —have long fascinated everyone from naturalists to artists • Dumb parts, properly connected into a swarm, yield smart results. • Using ants and other social insects as models, computer scientists have created software agents that cooperate to solve complex problems, such as the rerouting of traffic in a busy telecom network knowledge management, Calpe

4. What is swarm intelligence? • The term Swarm Intelligence (SI) was coined in the late 1980s • Social insects work without supervision. In fact, their teamwork is largely self-organized, and coordination arises from the different interactions among individuals in the colony. Although these interactions might be primitive (one ant merely following the trail left by another, for instance), taken together they result in efficient solutions to difficult problems (such as finding the shortest route to a food source among myriad possible paths). • The collective behavior that emerges from a group of social insects has been called “swarm intelligence”. knowledge management, Calpe

5. Some Definitions of Swarm Intelligence • Swarm Intelligence (SI) is the property of a system whereby the collective behaviors of unsophisticated agents interacting locally with their environment cause coherent functional global patterns to emerge. - Ramos, Fernandes et al. 2005 • Computational Swarm Intelligence (CSI) refers to algorithmic models. knowledge management, Calpe

6. “Swarm” = swarm, flock, herd, colony, gaggle, group, etc. • Any collection of agents where, if each agent enacts “simple” rules, the swarm exhibits a “complex” behavior. knowledge management, Calpe

7. Motivations of using SI • Dealing too complex problems: • Incapable to solve by human proposed solution • Absence of complete mathematical model • Existing of similar problems in nature: • Adaptation • Self-organization • Communication • Optimization • Characteristicsof a swarm: • Distributed, no central control or data source; • Limited communication • No (explicit) model of the environment; • Perception of environment (sensing) • Ability to react to environment changes knowledge management, Calpe

8. Characteristics of SI • Initially inspired by how social insects operate –shaped by millions of years of evolution. • Swarm Intelligence (SI) is the property of a system whereby the collective behaviors of (unsophisticated) agents interacting locally with their environment cause coherent functional global patterns to emerge. • It is a mindset rather than a technology. • It is a bottom-up approach to controlling and optimizing distributed systems • It use resilient, decentralized, self-organized techniques • It has limited communication • No (explicit) model of the environment • Perception of environment (sensing) • Ability to react to environment changes knowledge management, Calpe

9. The advantages of Swarm Intelligence (SI) • Flexibility: the group (swarm) can quickly respond to internal perturbations and external challenges. • Adaptability: The group can adapt to a changing environment. • Robustness: even if one or more individuals in the group fail, the group can still complete its tasks. • Self-organization: Paths to solutions are emergent rather than predefined. • Decentralized: the group needs relatively little supervision or top-down control. In other words, there is no central control(ler) in the colony. • Scalability: the control mechanisms used are not dependent on the number of agents in the swarm knowledge management, Calpe

10. Food source Nest 2. Foraging for solution • Ant foraging model: The way ants forage food holds valuable insights (Ant Colony Optimization) • Ants are able to find the shortest path form nest to a food source by laying and following chemical trails. • Individual ants emit a chemical substance – a pheromone – which then attracts other ants. • Probability of choosing a branch of a path at a certain time depends on the total amount of pheromone on the branch. • The choice is proportional to the number of ants that have used the branches. • Basic Rules: Lay pheromone and Follow the trails of others. knowledge management, Calpe

11. How does it function? Ants collectively select the shortest path to the food source. knowledge management, Calpe

12. Examples of Ant foraging models(Ant colony optimization in part III): Variations of this simple yet powerful approach can help solve a number of business problem: • Unpredictable environment of a telecommunication network • Effective cargo and vehicles routing. • The efficiency of factory scheduling. knowledge management, Calpe

13. Some Applications • Vehicle Routing Problem (VRP): The VRP is similar to the TSP, but is complicated by multiple vehicles, vehicle capacity, pick-up and drop off points (which can dictate vehicle packing and scheduling). Bernd Mullenheimer, Richard Hartl and Christine Strauss developed an Ant Colony algorithm for solving the VRP; and Pina Petroli truck routing • Scheduling : Scheduling is a widespread problem of practical importance. Paul Forsyth & Anthony Wren, University of Leeds Computer Science department developed a bus driver scheduling application using ant colony concepts. Air Liquide supply chain optimization and control; Unilever plant scheduling • Telecommunication Networks : Network routing refers to the activity of creating, maintaining and using routing tables (one for each node in the network) to determine where to direct an incoming data stream so that it can continue its travel through the network. In telecommunications, this is an extremely difficult problem because of the constant changes in network traffic load. The Ant Colony algorithm provides adaptive advantages that can adjust to traffic load. British Telecom, France Telecom, MCI routing in communications networks knowledge management, Calpe

14. 3. The task of dividing tasks • Honeybee model: The way insects allocate labor holds Valuable insights • Dividing tasks: In a honeybee colony, individuals specialize in certain tasks, and yet the allocation of work is very flexible. When food is scarce, nurse bees will help by foraging. knowledge management, Calpe

15. Case 1: Scheduling paint booths • In the factory, the booths must paint truck coming off an assembly line. When necessary, a booth can be change the color it’s using, but doing so is time-consuming and costly. • The booths can be thought of as honeybees governed by the following rule: • An individual performs the tasks for which it is specialized unless it perceive an important need to perform another function. • A booth with red paint will continue to handle orders of that color unless a job marked “urgent” requires a white truck and the queues at the other booths, particularly those specializing in white, are much longer knowledge management, Calpe

16. Case 1: Scheduling paint booths (cont.) • Although this basic rule sounds simplistic, in practice it is very effective. • It enables the paint booths to determine their own schedules with higher efficiency—specifically, fewer color changes—than a centralized computer can provide. • And the method is adept at responding to changes in consumer demand. If the number of trucks that need to be painted blue surges unexpectedly, other booths can quickly forgo their specialty colors to accommodate the unassigned vehicles. • Furthermore, the system copes easily with glitches. When a paint booth breaks down, other stations compensate swiftly by immediately divvying up the additional load. knowledge management, Calpe

17. Case 2: “Bucket brigade” model • Another useful model of work allocation comes from seed-harvester ants carrying food back to their nest. • Like runners transferring a baton in a relay race, the ants pass food down a chain. • But the ants are not stationary, and their transfer points are not fixed: an ant carries the food down the chain until it reaches the next ant, and after transferring the food, it turns back until it meets the previous ant in the chain to receive its next load. • The only fixed location in this operation are start (the food source) and the end (the nest).. knowledge management, Calpe

18. Case 2: “Bucket brigade” model (cont.) • The simple approach can dramatically increase the efficiency of operations in which work is passed from one person to another. For example, it can be applied to order pickers at a large distribution center of a major retail chain or to allocate workers in an product assembly line. • The warehouse used Zone approach, in which each worker was responsible for a particular part of the order, and next person could not begin until the first person complete that task. • One problem with zone approaches is the wide variation in the rates at which different employees work. • A supervisor had to monitor each aisle to correct the congestions that inevitably occurred. knowledge management, Calpe

19. Case 2: “Bucket brigade” model (cont.) • Set up a new simple rule: “Continue picking out products to fill the order until the person downstream from you takes over you work; then head upstram to take over the next person’s work” • The optimum sequence of workers is from the slowest to the fastest. • “Bucket brigade” model allows a work line to balance itself – that is, the optimum solution emerges without any intervention by managers • In assembly line, it can function as a self-organizing system that spontaneously achieves its own optimum configuration, without special equipment, time-motion studies, work-content models, management, or software control systems. knowledge management, Calpe

20. 4. Simple rules rule • The most powerful-and fascinating-insight from swarm intelligence is that complex collective behavior can emerge from individuals following simple rules. • For social insects, millions of years of evolution have fine-tuned those rules for great efficiency , flexibility , and robustness. Can managers develop similar rules to shape the behavior of their organizations and replace rigid command-and-control structures? knowledge management, Calpe

21. An Example: Bird flocking • “Boids” model: “bird-oid” objects (also schooling fish) • Model: biologically and physically sound * Individual has only local knowledge * Has certain cognitive capabilities * Is bound by the law of physic knowledge management, Calpe

22. Three rules • Collision avoidance (avoid collisions with neighboring boids) • Velocity Matching (match the velocity of neighbouring boids) • Flock centering (stay near neighnoring boids) knowledge management, Calpe

23. Bad news, good news • Bad news • Difficult to predict collective behavior from individual rules. • Interrogate one of the participants, it won’t tell you anything about the function of the group. • Small changes in rules lead to different group-level behavior. • Individual behavior looks like noise: how do you detect threats? • Good news • Possible to efficiently control organization or manipulate groups using simple rules. • Possible to predict group-level outcome using bottom simulation. knowledge management, Calpe

24. Simple rules modeling at Southwest Airlines • Problem • Optimize cargo routing • Use simple rules • Results • 71% improvement • At least $10m/yr knowledge management, Calpe

25. Challenges in simple rules modeling • Important research thrust: If we want the swarm to exhibit a certain behavior (accomplish an objective), what simple rules should the agents follow? • Have to know relationship between rules and behavior knowledge management, Calpe

26. 5. Raiding new markets • Laying pheromone is a form of “mass recruitment”; • In some species, though, an ant that finds a food source returns to the nest and vibrates its antennae to convince one other nest mate to return to the site. That’s “tandem recruitment”; • In other cases, an ant vibrates its antennae to get a number of nest mates to follow. That’s “group recruitment”. knowledge management, Calpe

27. Raiding new markets (cont.) • Mass recruitment is most often associated with large colonies • Tandem recruitment with small colonies • Group recruitment with medium-sized ones Example: Louise Kitchen, 1999, Thank s to Group recruitment, that new “food source” handles approximately $1 billion of transactions daily and has added a few billion dollars to Enron’s market capitalization. knowledge management, Calpe

28. The right nurturing environment for group recruitment • Maintain their ability to explore new opportunities while exploiting existing ones; • Enable a person with an idea to recruit others; • Allow, but not force, people to be recruited, even when they are working in a core business; • Let the system self-select the best ideas; and • Support the winning ideas with sufficient resources. knowledge management, Calpe

29. 6. A swarm of possibilities • The possible applications of swarm intelligence may be limited only by the imagination. • The way insects cluster their colony’s dead and sort their larvae has led to a novel approach for banks to use to analyze their data for interesting commonalities among customers. • Reconfigurable robots swarms can assemble themselves into vacuum cleaners and other home appliances or move an object collaboratively. • More and more. knowledge management, Calpe

30. Some future studies • When a honeybee colony becomes too large - that is ,when it reaches a point of diminishing returns - the nest splits into two; exactly what rules bees follow to do this remains a mystery. • Such knowledge maybe inspired: When should a large corporations determine to spin off some of their operations? • A queen wasp, fearing that the departure of some of her subordinates could cripple the colony, induces them to stay by granting them the right to lay eggs. The amount of this “staying incentive” depends on ecological conditions. If, say, the weather is mild and food abundant, the queen must offer greater inducements. • The parallel with managers trying to retain top talent in a booming economy is striking. • The parallel between social insects and people are more than just conceptual: they can have practical and useful significance, as recent research has shown. knowledge management, Calpe

31. 7. Conclusion • SW is becoming a valuable tool for optimizing the operations of various businesses. • Through SI provides a fresh new framework for solving such problems, and it still questions the wisdom of certain assumptions regarding the need for employee supervision through command-and control management. • In the future, some companies could build their entire businesses from the ground up using the principles of swarm intelligence, integrating the approach throughout their operations, organization, and strategy. • The result: the ultimate self-organizing enterprise that could adapt quickly and instinctively to fast changing markets. knowledge management, Calpe

32. Thanks for Your Attention! knowledge management, Calpe