Hybrid differential evolution/estimation of distribution algorithm based on adaptive incremental learning

Estimation of Distribution Algorithm (EDA) can describe the relationship between the variables with its probabilistic model, which has the good global searching ability. But its convergence speed is too fast, so it is easy to trap in the local optimum, while Differential Evolution (DE) has the good capability of local searching. In response to these issues, a hybrid DE/EDA based on adaptive incremental learning is proposed in this paper, which introduces the factor of the excellent population based on adaptive incremental learning to adjust these tow algorithms with an adaptive strategy, and accelerates the convergence speed. Meantime, the Markov chain is established for this algorithm and its global convergence is also proved with the theory of stochastic processes. The experimental results on eight benchmark functions demonstrate its global convergence. 1553-9105/Copyright © 2014 Binary Information Press.