Heuristic Design

Heuristic design aims to automate the creation of efficient algorithms for solving complex optimization problems, traditionally a laborious manual process. Current research heavily utilizes large language models (LLMs) integrated with evolutionary algorithms and symbolic regression to generate and refine heuristics, often employing multi-objective optimization to balance performance, efficiency, and scalability. This automated approach promises to significantly accelerate algorithm development across diverse fields, from combinatorial optimization to AI planning, by offering a more efficient and potentially more innovative alternative to manual design. The resulting algorithms are showing competitive or superior performance compared to existing methods in various benchmark problems.