Hyper heuristics for scheduling problems

[Master's thesis]

In many application areas of scheduling exact algorithms (i.e. algorithms that lead to a guaranteed optimal solution) are out of scope. The main reason is that such algorithms would run too long, for example years, when we are today interested in the result. In such situtation heuristics are used: heuristics are methods that generate solutions to the problem under investigation, but (usually) it is unknown if the generated result is optimal.

Usually one can think of many different heuristics, all with their own parameters. Once a set of heuristics is developed, an interesting question is what is the best combination of these heuristics. To answer this question, hyper heuristics can be used. So, hyper heuristics try to generate a combination of basic heuristics that works the best in a problem area. These areas could be timetabling, vehicle routing, packing problems, etc.

For more information, please contact Gerhard Post.