Distribution Algorithm

Estimation of distribution algorithms (EDAs) are a class of evolutionary optimization algorithms that build probabilistic models of promising solutions to guide the search for optima, rather than relying solely on traditional genetic operators. Current research focuses on improving EDA efficiency and effectiveness for various problem types, including those involving permutations, assignments, and graph coloring, exploring model architectures like doubly stochastic matrices and multivariate models alongside simpler univariate approaches. EDAs show promise in diverse applications, from semiconductor layout optimization and cancer chemotherapy planning to solving complex combinatorial problems, with ongoing efforts to better understand their performance on challenging landscapes and develop metrics for predicting problem difficulty.