Swarm computing is based on analogies with social insects such as ants and bees. These insects behave in intelligent ways as a collective even though each individual insect is very simple. In swarm computing many simple individual agents are prodused which explore a spece, for example to find an optmal solution. They operate semi-independently with a level of simple communication that nudges their behaviour. For examaple, if an agent in the swarm has found a good solution it may retrace the intermediate states it took to get there labelling them as good, but with decaying strengh, thus mirroring phermone trails. Swarm computing is similar to {[ensemble method}}, but the latter are more about forming a diverse group of simple {[decsions making}} agents that together are used repeatedly to solve similar problems, wheres in swarm computing the agents are near identical and it usualy only the final result that matters.
Used on page 535