Tabu search
Tabu search is a metaheuristic to avoid getting stuck in local optimum during a neighborhood search.
For example, if you use the basic rule of the "first improvement" among the neighbours, there is chances you're missing potential good solutions that need to go through a temporary none-improving path.
Tabu search impose one or many rules to prohibit movements and relax the basic rule.
- Maximum iterations (stopping criteria)
- No inverse movements
- Avoid doing
time the same movement - Force movement that never occurs
- …, etc.
Problems
Here is some problems you can apply tabu search to.