Distributed intelligence

The distributed intelligence indicates the pooling of the capacity of reasoning of several Individu S in order to achieve a goal which it is impossible or more difficult to solve without this co-operation.

This concept was not born from the Artificial intelligence, in spite of the share which he plays in this field. The computer programs with distributed Artificial intelligence exist, but they owe their existence with a similar concept of the Nature, like much of human inventions.

The most obvious case is the colony of Fourmi S. When an object (a stone, for example) appears on the usual way of a colony of ants, its individuals must find the means simplest (and often fastest) to circumvent this new obstacle. The ants have neither the Intelligence, nor the size necessary to know if it is better to pass on the right or on the left. The first ants choose a side then randomly. When these ants arrive on other side, the ants coming from opposite find Statistique lies more ants coming on the shortest side (time of way being less long), and thanks to a odorous marker, chooses this way.

Distributed artificial intelligence and intelligence in swarm: collective intelligence of simple groups of agents

The term “intelligence in swarm” (in English “ swarm intelligence ”), was stated in the context of artificial systems by Gerardo Beni in 1989 ( Proceedings off the Seventh Annual Meeting off the Robotics Society off Japan ).

The concept was then defined in the following way: “ Swarm Intelligence has off property systems off not-intelligent robots exhibiting collectively intelligent behavior . ”. In other words, “the intelligence in swarm is a property of systems of not-intelligent robots which show an intelligent behavior collectively”.

Central brain with the intelligence in swarm - To reinvent software coding

In an ambient network, the processors and their software are diffused in the objects beyond the conventional concept of our current computers. These potentially tiny objects, will have to be able to exploit great quantities of data, little structured and all that in a context of less and less fixed connections - hyper-mobility obliges even not very sure.

These information systems require an new approach of the programming on which new constraints weigh (extreme variety of the equipment, mobility, format of data radically new, degraded operations of connections, obligation of certification of the software,…).

Bases of the codes of tomorrow, i.e. the programs of the applications of distributed intelligence rest on three principles:

the “coupling” or the reciprocal interaction between the codes of two objects is all the more loose as the number of objects is high, the asynchronous means of communication is thus probably the future of software in swarm.
concept of achievable microcomponents, indissociable of this ordered propagation with very large scales of the code.
in an ultimate phase, the algorithms should adapt to the problems arising, i.e. to develop themselves of new methods to solve them: the software of the future will evolve/move according to their environment. It is the exploratory concept of mutant application.

Random links: