Tabu search is an optimization heuristic that uses a memory to guide the search into unexplored regions.
Tabu search is a trajectory-based metaheuristic that iteratively modifies a solution through small changes. It evaluates each of a possible number of changes and in general chooses that change that maximizes the gain on the fitness function. However, it also remembers previously made changes and avoids to revert changes that were made in the recent history, thus preventing the search to return to already explored areas. More advanced versions also feature additional memories that influence the decision to move towards promising areas.
