Make your own free website on Tripod.com

>Introduction >Background and Method >Demo >Application >Comparison with GA >Artificial Intelligence > References

 

Applications of Swarm Intelligence

The idea of swarm intelligence has been successfully used in :

Functional optimization , Artificial Neural Network training , Fuzzy System Control , and artificial life . Both continuous and discrete optimization can be solved using the idea , in appropriate form .

Using ants , and other social insects as an example , scientists have formulated algorithms even for the rerouting of traffic in a busy telecom network . Similary , we can coordinate the behaviour of teams of autonomous robotic agents to cooperate , using swarm behaviour . Some examples of the variety of work done using Swarm Intelligence :

Routing Algorithms , Business Strategies , simulation Social Behaviour , optimization in Power Systems Engineering

I have conducted an experimented with principles of Swarm Intelligence , on teams of ( simulated ) mobile robots . The IBM Robocode simulator was used . Various observations were made , for asynchronous teams of agents using Swarm Intelligence . A slight modification of the definition of 'gbest' and 'pbest' was used for implementation purposes , because in this case there was no precise function of the form f ( x , y , z... ) to be optimized . Rather , we had to analyse the kind of cooperation which developed in the agents . The report is available here : swarmai.html

Also , Patricle Swarm Optimization ( PSO ) can be extended for the following tasks as well :

1 ) DISCRETE OPTIMIZATION :

The idea of PSO can be extended to problems of Combinatorial Optimization as well . Kennedy and Eberhart developed a discret binary version . In their model , the probability of an agent's deciding yes or no , true or false , is a function of personal and social factors .

P ( s = 1 ) = f ( s , v , pbest , gbest )

The activation function is the sigmoid function ( 1 / ( 1 + exp ( -v) ) ) .

 

 

 

 

 

 

 

 

>Introduction >Background and Method >Demo >Application >Comparison with GA >Artificial Intelligence > References