Posted by: zyxo | August 10, 2008

Swarm information transfer techniques

Fanning honeybee exposes Nasonov gland (white-...Image via WikipediaSwarms are more intelligent than the individuals of the swarm. How could for example termites otherwise build their sophisticated termite hills ? Or how could ants show such complex behaviour ?
A good article to learn more about it is this from Eric Bonabeau, Marco Dorigo and Guy Theraulaz or this presentation from Thiemo Krink.

This intelligence is the result of information that is transferred to the other members of the swarm. This information transfer can be done two different ways :

  1. stigmergy : indirect communication based on modification of the environment. For example ants, who deposit a trail of pheromones on their way, to lead the other ants to the food source
  2. direct communication : honeybees who exhibit a particular dance to show the direction and distance of the food source to the other bees.

In artificial swarm intelligence several algorithms have been developed to copy this swarm intelligence mechanisms for data mining purposes.

Here I want to go a little further than just ants and honeybees and see if other swarm animals can offer some ways to develop swarm intelligence algorithms.

I already mentioned ants and honeybees.
In a post a while ago I also wrote that one of the algorithms used in the experimental software Antminer+ more mimics the behaviour of locusts than that of ants.
If we zoom in on locusts, how could these animals transfer their information to other locusts ? By stigmergy. As ants only deposit pheromone to indicate the path, artificial locusts (not real ones) have to do better than that : as locusts jump from one place to the other, the previous locust has to indicate the direction where to jump and eventually the distance of the jump, in order to get to a good spot. This complicates the life of the locusts a bit, because they have to come back each time to leave their information in the previous spot. But it could work.
So you can see that locusts and honeybees sort of communicate the address of the good spot to their swarm fellows.
There is yet another animal species that does this in a even more accurate way : homo sapiens. We, humans.
So why not build a swarm intelligence algorithm where people come back from, say a shopping trip, and write down the address of the store where it is cheap, or good quality, or whatever is favorable.
As in any swarm intelligence algorithm there is some incertainty about all this information transfer, so that the others arrive approximately on the spot, but eventually also look in the neighbourhood.

And just another thought : why in stead of only communicating info of the favorable spots, should we not also communicate info of the particularly bad spots, in order to get others to avoid those spots and do not waste any time there when they do some random searching?

Zemanta Pixie

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s


%d bloggers like this: