Reducible intelligence: Swarm Intelligence

| 13 Comments

The National Geographic shows how intelligence can be reduced to simpler processes and rules. The work is important as it shows, contrary to ID’s assertions, that intelligence can in fact be reducible to processes of regularity and chance. From ants to bees to fish to caribous, the researchers found that “swarm intelligence” is based on simple creatures following simple rules based only on local information. In other words, none of the creatures sees the bigger picture and yet the swarm “responds” in an intelligent way.

Several researchers have used the concept of swarm intelligence to find solutions to similar problems. For instance, Southwest Airlines used the principle to find an optimal schedule for its planes leaving and arriving at gates. In another instance a company optimized its profits by scheduling which plants were to deliver to which customers.

WIkipedia explains:

SI systems are typically made up of a population of simple agents interacting locally with one another and with their environment. Although there is normally no centralized control structure dictating how individual agents should behave, local interactions between such agents often lead to the emergence of global behavior. Examples of systems like this can be found in nature, including ant colonies, bird flocking, animal herding, bacterial growth, and fish schooling.

Particle Swarm Optimization (PSO) is similar to evolutionary computation techniques:

PSO shares many similarities with evolutionary computation techniques such as Genetic Algorithms (GA). The system is initialized with a population of random solutions and searches for optima by updating generations. However, unlike GA, PSO has no evolution operators such as crossover and mutation. In PSO, the potential solutions, called particles, fly through the problem space by following the current optimum particles.

Two popular approaches include Ant Colony Optimization and particles swarm optimization.

As I explained, there are some similarities between PSO and ACO and genetic algorithms but PSO’s have a significant difference

Compared with genetic algorithms (GAs), the information sharing mechanism in PSO is significantly different. In GAs, chromosomes share information with each other. So the whole population moves like a one group towards an optimal area. In PSO, only gBest (or lBest) gives out the information to others. It is a one -way information sharing mechanism. The evolution only looks for the best solution. Compared with GA, all the particles tend to converge to the best solution quickly even in the local version in most cases.

Swarm optimization is basically an example of self organization where complexity arises out of the combined interactions of local, simple rules. Such emergence shows how ID’s position on intelligence is at least contradicted at the level of swarm intelligence.

It occurred to me that the stress response, found in many organisms, which was once seen as evidence of directed mutations, may be another example of parallel computing and problem solving. By increasing the mutation rates, the bacteria can explore a larger phenotype space and find a potential solution for the stressor. Once a solution has been found, the genetic material can be shared quickly via horizontal gene transfer. Foraging behavior of bacteria also seem to be a good example of swarm intelligence.

13 Comments

An interesting finding

 Swarming consits of coordination of movement, but when a (food) source is reached, competition may prevail. From competitive interactions (dominance interactions) a dominance hierarchy arises. Both in animals and in humans, the effects of victory and defeat in competitive interactions are self-reinforcing: losing/winning an interaction (or match) increases the chance to lose/win the next one. This is called the winner/loser effect (Chase, 1974; Chase et al., 1994). Hemelrijk (Chapter 5) shows in an agent-based model that the tendency to group in combination with such winner/loser effect leads to many emergent patterns of behaviour. For instance, at a high intensity of aggression, a steep hierarchy develops and also a spatial structure with dominant individuals in the centre and subordinates at the periphery. Both reinforce each other and lead to further emergent phenomena, which resemble those found in certain species of primates. It appears that increasing only one parameter in the model (intensity of aggression) causes a switch in the artificial society from characteristics typical of an egalitarian society to those of a despotic one as known from macaques. Thus, many different characteristics of societies may arise from a change in a single internal trait. It follows that, obviously, the genetic inheritance of a social system is then only partial and characteristics of the social system are largely formed through the interactions among the individuals.

source

Simple rules leading to complex social behaviors, wow.

Ah-ha, this is it. This is the break that ID has been looking for. God is a swarm. That explains everything. No need for any more research, we can all go home.

With all respect to the field of swarm intelligence, but on the face of it I think the comparison with genetic algorithms mechanisms has been a little forced, at least in the reference given in the post.

Swarm intelligences are glorified particles with the complexity of near-local or non-local interaction (information transfer) between particle agents. The natural connection is instead between its respective uses and characteristics in say biology and problem solving. In the later case a difference to physics may be made by emulating instantaneous information transfer.

I’m sure everyone has their favorite example of organization from simple processes, whether simple physics or not. Crystallization, phase changes or biological swarms are often mentioned. My own is particle piles. Sand piles self-organize towards an, IIRC, universal angle of slope by local avalanches when approaching it.

This self-organized criticality depends, again assuming IIRC, in principle only on gravitational strength and not in a first approximation on particle size or static friction or fluid viscosity. Looking at a pile in the solar system, you could then in principle locate on which body it formed.

Similar self-organization by local processes is seen in wind-transported sand dunes, irrespective of wind strength. If one studies the Titan’s dune fields compared to Earth’s dune fields I’m pretty sure one will again extract different values of universal traits.

I’m also pretty sure one can wax endlessly philosophically over universal and/or aggregated phenomena. “Unity in numbers”, “The more things change, the more they stay the same”, … :-P

“wax endlessly philosophically” - wax endlessly philosophical.

One quote in particular seems relevant to the ID debate:

The bees’ rules for decision-making—seek a diversity of options, encourage a free competition among ideas, and use an effective mechanism to narrow choices—

The “teach the controversy” approach seems to emphasize the first two points. But it is science that applies the third. It seems that many ID proponents want to emphasize the diversity of opinion without emphasizing the mechanism for sorting different opinions and weighing them against facts.

WAX PHILOSOPHICAL

… now there’s a good name for a beauty salon

source: “Simple rules leading to complex social behaviors, wow.”

Physicist Stephen Wolfram of Mathematica [ http://www.wolfram.com/ ] have been trying to find simple rules which brings out complex things. His blog http://blog.wolfram.com/2007/09/my_[…]ml?lid=title gives a good overview of his approach.

No one mentioned free market economics as an example of swarm intelligence yet. In terms of understanding I think it is at least somewhat beneficial to take a broad view and see all of these phenomena under the immergent properties description. Simples rules and many interactions can lead to very complex behaviors.

The National Geographic shows how intelligence can be reduced to simpler processes and rules. The work is important as it shows, contrary to ID’s assertions, that intelligence can in fact be reducible to processes of regularity and chance.

This sort of work is also more than a decade old. Stephen Budiansky wrote at length about this view of intelligence in his 1998 book If A Lion Could Talk: Animal Intelligence and the Evolution of Consciousness.

WAX PHILOSOPHICAL

… now there’s a good name for a beauty salon

:-)

I guess then WANE SCIENTIFICALLY

… is a good description of creationism.

God is a swarm

Explaining why Genesis 1 uses a word for God (Elohim) that is morphologically plural but often functionally singular.

No one mentioned free market economics as an example of swarm intelligence yet

My consultancy experience gives me the distinct impression that an awful lot of large companies work this way, especially those in rapidly changing markets. No-one has much idea what is really going on (not even, contrary to their claims, the board members) but somehow products get made and surprisingly frequently they even work.

My consultancy experience gives me the distinct impression that an awful lot of large companies work this way, especially those in rapidly changing markets. No-one has much idea what is really going on (not even, contrary to their claims, the board members) but somehow products get made and surprisingly frequently they even work.

I used to work in the research labs of a well-known major corporation what was run by mob intelligence. That company is now on the rocks. There seems to be a big difference between mob and swarm effects.

About this Entry

This page contains a single entry by PvM published on October 22, 2007 10:09 PM.

Science v Intelligent Design: ID and whales was the previous entry in this blog.

Ben Stein v Intelligent Design: Filling in the gaps is the next entry in this blog.

Find recent content on the main index or look in the archives to find all content.

Categories

Archives

Author Archives

Powered by Movable Type 4.361

Site Meter