On the convergence of a factorized distribution algorithm with truncation selection

nical Abstract Optimization is to find the "best" solution to a problem where the quality of a solution can be measured by a given criterion. Estimation of Distribution Algorithms (EDA) generate a sequence of populations (collections) of candidate solutions for solving an optimization problem. They evaluate and analyze each member in the current population, and based on the results estimate the probability of each possible solution being the best. Then a population of new candidate solutions are generated according to these estimated probabilities. EDAs have become an increasing popular optimization tool. However, it is still unclear in theory when EDAs work. In this paper, we prove that a Fractorized Distribution Algorithm (FDA) with truncation selection, an instance of EDAs, can solve a class of optimization problems. This work was supported by EPSRC under Grant GR/R64742/01 1 

• Type: Journal
• Year: 2004
• Where: COMPLEXITY
• Authors: Qingfu Zhang
• Keywords: COMPLEXITY 2004, Distribution Algorithm, Nical Abstract Optimization, Solutions