Search

Differential evolution with a multi-population based ensemble of mutation strategies (MPEDE) has been considered among the most efficient Evolutionary Algorithms for global optimization. Our research results reveal that the performance of MPEDE may be improved by adding an information sharing mechanism and modifying the grouping mechanism. In MPEDE, the entire population is divided into four subpopulations, and most computing resources are allocated to the best strategy, but a better strategy has the same computing resources as the worst strategy. In order to rationally distribute computational resources, a differential evolution variant with multi-population cooperation and multi-strategy integration (MPMSDE) is proposed in this paper. MPMSDE develops a new grouping method instead of the grouping method in MPEDE, and the new grouping method utilizes the ranking of strategies to assign computational resources to different strategies. Also, an information sharing mechanism is introduced in the largest sub-population to avoid falling into local optimum. In MPMSDE, a new mutation strategy, "DE/pbad-to-pbest-to-gbest/1" , is used to replace the mutation strategy "DE/rand/1" in MPEDE. The new strategy not only uses personal history optimal solution and the worst solution but also uses the global best solution to update individuals. The new strategy can not only balance exploration and exploitation but also can accelerate the convergence of the algorithm. The performance of MPMSDE is compared with MPEDE and other state-of-the-art evolutionary algorithms on CEC2005 and CEC2014 benchmark functions. The experimental results show that the performance of the MPMSDE algorithm is very competitive in calculation accuracy and convergence speed. [ABSTRACT FROM AUTHOR]

Abstract: Intelligent evolutionary algorithms have been widely used to solve large-scale, complex global optimization problems. Co-evolutionary algorithm (CEA), cultural algorithm (CA), and particle swarm optimization (PSO) are all promising methods in the field of intelligent computation. In this paper, a hybrid co-evolutionary cultural algorithm based on particle swarm optimization (CECBPSO) is proposed. In CECBPSO, a novel space called shared global belief space (SGBS) is introduced into the co-evolutionary mechanism, and a new co-evolutionary cultural framework is built. Through the synergistic mechanism, the algorithm has higher probability of avoiding local optima and the whole swarm can find global optima more quickly. Factorial Design (FD) approach is used in this paper in order to get a guideline on how to tune the designed parameters in CECBPSO. Extensive computational studies are also carried out to evaluate the performance of CECBPSO on thirteen benchmark functions and three real-life optimization problems. The results show that the proposed algorithm has superior performance to other compared algorithms in terms of accuracy and convergence speed, especially on high-dimensional problems. [Copyright &y& Elsevier]

Abstract: Recently, Gandomi and Alavi proposed a meta-heuristic optimization algorithm, called Krill Herd (KH), for global optimization [Gandomi AH, Alavi AH. Krill Herd: A New Bio-Inspired Optimization Algorithm. Communications in Nonlinear Science and Numerical Simulation, 17(12), 4831–4845, 2012.]. This paper represents an optimization method to global optimization using a novel variant of KH. This method is called the Stud Krill Herd (SKH). Similar to genetic reproduction mechanisms added to KH method, an updated genetic reproduction schemes, called stud selection and crossover (SSC) operator, is introduced into the KH during the krill updating process dealing with numerical optimization problems. The introduced SSC operator is originated from original Stud genetic algorithm. In SSC operator, the best krill, the Stud, provides its optimal information for all the other individuals in the population using general genetic operators instead of stochastic selection. This approach appears to be well capable of solving various functions. Several problems are used to test the SKH method. In addition, the influence of the different crossover types on convergence and performance is carefully studied. Experimental results indicate an instructive addition to the portfolio of swarm intelligence techniques. [Copyright &y& Elsevier]

Abstract: Differential Evolution (DE) is a simple yet efficient global optimization algorithm. However, the standard DE and most of its variants operate in the continuous space, which cannot solve the binary-coded optimization problems directly. To tackle this problem, this paper proposes a novel modified binary differential evolution algorithm (NMBDE) inspired by the concept of Estimation of Distribution Algorithm and DE. A novel probability estimation operator is developed for NMBDE, which can efficiently maintain diversity of population and achieve a better tradeoff between the exploration and exploitation capabilities by cooperating with the selection operator. Furthermore, the parameter study of NMBDE is run and the analysis is performed to improve the global search ability and scalability of algorithm. The effectiveness and efficiency of NMBDE was verified in applications to the numerical optimization and multidimensional knapsack problems. The experimental results demonstrate that NMBDE has the better global search ability and outperforms the discrete binary DE, the modified binary DE, the discrete binary Particle Swarm Optimization and the binary Ant System in terms of accuracy and convergence speed. [Copyright &y& Elsevier]

Abstract: This paper presents a new global optimal independent component analysis (ICA) algorithm. The constrained optimization problem, which is typically encountered with conventional ICA, is converted to an unconstrained problem with the orthogonal projection approach. Then the particle swarm optimization (PSO) is employed to solve the unconstrained problem and determine the separating matrix of ICA. Applications in the analysis of the magnetoencephalographic recordings (MEG) illustrate the efficiency of the proposed PSO–ICA approach. [Copyright &y& Elsevier]

In this paper, we present meta-learning evolutionary artificial neural network (MLEANN), an automatic computational framework for the adaptive optimization of artificial neural networks (ANNs) wherein the neural network architecture, activation function, connection weights; learning algorithm and its parameters are adapted according to the problem. We explored the performance of MLEANN and conventionally designed ANNs for function approximation problems. To evaluate the comparative performance, we used three different well-known chaotic time series. We also present the state-of-the-art popular neural network learning algorithms and some experimentation results related to convergence speed and generalization performance. We explored the performance of backpropagation algorithm; conjugate gradient algorithm, quasi-Newton algorithm and Levenberg–Marquardt algorithm for the three chaotic time series. Performances of the different learning algorithms were evaluated when the activation functions and architecture were changed. We further present the theoretical background, algorithm, design strategy and further demonstrate how effective and inevitable is the proposed MLEANN framework to design a neural network, which is smaller, faster and with a better generalization performance. [Copyright &y& Elsevier]