Heuristics and Search Methods

Description

This course discusses the most recent developments in the area of non-exhaustive search methods for large and complex search spaces. For most optimization problems encountered in practice, the search space is not convex and contains far too many candidate solutions to enumerate them all in order to find the optimal solution. That is why there is a great need for search methods that crawl through the search space in a more intuitive way, converging very fast to solutions which, although perhaps suboptimal, are still very good. Often, the success of such heuristic methods depends on whether one succeeds in implementing problem-specific knowledge into the search method. In the course, several classes of heuristic approaches are discussed which prove to be extraordinarily successful with some of the hardest realistic problems.

Contents

• Local search methods
• Stochastic local search
• Constraint programming
• Neural networks
• Principles of tabu search
• Genetic algorithms
• Simulated annealing