Simulated Bee Colony Algorithms

In the April 2011 issue of MSDN Magazine I wrote an article which describes Simulated Bee Colony (SBC) algorithms. See SBC algorithms model the behavior of foraging honey bees to find good (but not necessarily optimal) solutions to combinatorial optimization problems which cannot be solved using normal techniques. In the article I use an SBC to solve the Traveling Salesman Problem where you have a set of cities and must visit each city exactly once in such a way that the total distance traveled is minimized. See the screenshot below. SBC algorithms are related to, but distinct from, other meta-heuristics based on the behavior of natural systems such as Genetic Algorithms based on chromosomes and mating, Ant Colony Optimization based on ant pheromones, and Simulated Annealing based on cooling metal. After I wrote the MSDN Magazine article I applied an SBC algorithm to the problem of graph partitioning and was able to find several new world-record best partitionings for benchmark problems.

This entry was posted in Machine Learning. Bookmark the permalink.