Next: Complexity estimation Up: The genetic algorithm Previous: The social aspect of   Contents   Index

# Why genetic algorithms are good for encoding multimedia objects

There is, in general, no algorithm for converting raster graphics directly into memory-friendly vector images, where the strengths of the vector description language are actually used. For the conversion from one multimedia format into another, there is often a similar problem, especially when in the first multimedia format, the properties of points of a euclidean (discrete) space are given directly (for example, to be able to store images directly) and the second multimedia format encodes the multimedia objects with complex objects (such as rectangles and / or circles).

Since a genetic algorithm has the potential to generate all possible descriptions in a multimedia language and among these, of course, also good descriptions will be, that use the strengths of multimedia language in respect of the original object, a genetic algorithm has also the potential to generate good descriptions. If existing knowledge was well implemented into genetic algorithm, it can find good descriptions probably faster than a pure random search.

Yet another advantage of genetic algorithms is the great freedom in the choice of problem description (multimedia language) and the possible operators on it. This allows complete freedom to design a multimedia language according to your own ideas, wich has certain properties, such as readability or simplicity. Into the operators arbitrarily much knowledge can be incorporated. It is, for example, possible to use known good algorithms (e. g. to translate raster images into vector images) or parts of them in operators, so that the result from the genetic algorithm (on raster images) is at least as good as the algorithm, but even better results can be generate.

The major disadvantage of genetic algorithms that they need very much time or computing time is weakened by the fact, that this high initial cost can be "cheap" and pays off later. The genetic algorithm can run, for example, as a background process with low priority, so that it consumes superfluous processing power. Later, with the result that it has supplied, much bandwidth can be saved.

This all suggests to use genetic algorithms to encoding multimedia objects.

Next: Complexity estimation Up: The genetic algorithm Previous: The social aspect of   Contents   Index
Betti Ã–sterholz 2013-02-13