Nonlinear system identification using memetic differential evolution trained neural networks

Several gradient-based approaches such as back propagation (BP) and Levenberg Marquardt (LM) methods have been developed for training the neural network (NN) based systems. But, for multimodal cost functions these procedures may lead to local minima, therefore, the evolutionary algorithms (EAs) based procedures are considered as promising alternatives. In this paper we focus on a memetic algorithm based approach for training the multilayer perceptron NN applied to nonlinear system identification. The proposed memetic algorithm is an alternative to gradient search methods, such as back-propagation and back-propagation with momentum which has inherent limitations of many local optima. Here we have proposed the identification of a nonlinear system using memetic differential evolution (DE) algorithm and compared the results with other six algorithms such as Back-propagation (BP), Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Differential Evolution (DE), Genetic Algorithm Back-propagation (GABP), Particle Swarm Optimization combined with Back-propagation (PSOBP). In the proposed system identification scheme, we have exploited DE to be hybridized with the back propagation algorithm, i.e. differential evolution back-propagation (DEBP) where the local search BP algorithm is used as an operator to DE. These algorithms have been tested on a standard benchmark problem for nonlinear system identification to prove their efficacy. First examples shows the comparison of different algorithms which proves that the proposed DEBP is having better identification capability in comparison to other. In example 2 good behavior of the identification method is tested on an one degree of freedom (1DOF) experimental aerodynamic test rig, a twin rotor multi-input-multi-output system (TRMS), finally it is applied to Box and Jenkins Gas furnace benchmark identification problem and its efficacy has been tested through correlation analysis.     

Group Search Optimizer: An Optimization Algorithm Inspired by Animal Searching Behavior

Nature-inspired optimization algorithms, notably evolutionary algorithms (EAs), have been widely used to solve various scientific and engineering problems because of to their simplicity and flexibility. Here we report a novel optimization algorithm, group search optimizer (GSO), which is inspired by animal behavior, especially animal searching behavior. The framework is mainly based on the producer-scrounger model, which assumes that group members search either for ldquofindingrdquo (producer) or for ldquojoiningrdquo (scrounger) opportunities. Based on this framework, concepts from animal searching behavior, e.g., animal scanning mechanisms, are employed metaphorically to design optimum searching strategies for solving continuous optimization problems. When tested against benchmark functions, in low and high dimensions, the GSO algorithm has competitive performance to other EAs in terms of accuracy and convergence speed, especially on high-dimensional multimodal problems. The GSO algorithm is also applied to train artificial neural networks. The promising results on three real-world benchmark problems show the applicability of GSO for problem solving.     

A new evolutionary approach to cutting stock problems with and without contiguity

Evolutionary algorithms (EAs) have been applied to many optimization problems successfully in recent years. The genetic algorithm (GAs) and evolutionary programming (EP) are two different types of EAs. GAs use crossover as the primary search operator and mutation as a background operator, while EP uses mutation as the primary search operator and does not employ any crossover. This paper proposes a novel EP algorithm for cutting stock problems with and without contiguity. Two new mutation operators are proposed. Experimental studies have been carried out to examine the effectiveness of the EP algorithm. They show that EP can provide a simple yet more effective alternative to GAs in solving cutting stock problems with and without contiguity. The solutions found by EP are significantly better (in most cases) than or comparable to those found by GAs.Scope and purposeThe one-dimensional cutting stock problem (CSP) is one of the classical combinatorial optimization problems. While most previous work only considered minimizing trim loss, this paper considers CSPs with two objectives. One is the minimization of trim loss (i.e., wastage). The other is the minimization of the number of stocks with wastage, or the number of partially finished items (pattern sequencing or contiguity problem). Although some traditional OR techniques (e.g., programming based approaches) can find the global optimum for small CSPs, they are impractical to find the exact global optimum for large problems due to combinatorial explosion. Heuristic techniques (such as various hill-climbing algorithms) need to be used for large CSPs. One of the heuristic algorithms which have been applied to CSPs recently with success is the genetic algorithm (GA). This paper proposes a much simpler evolutionary algorithm than the GA, based on evolutionary programming (EP). The EP algorithm has been shown to perform significantly better than the GA for most benchmark problems we used and to be comparable to the GA for other problems.     

Evolutionary artificial neural networks: a review

This paper reviews the use of evolutionary algorithms (EAs) to optimize artificial neural networks (ANNs). First, we briefly introduce the basic principles of artificial neural networks and evolutionary algorithms and, by analyzing the advantages and disadvantages of EAs and ANNs, explain the advantages of using EAs to optimize ANNs. We then provide a brief survey on the basic theories and algorithms for optimizing the weights, optimizing the network architecture and optimizing the learning rules, and discuss recent research from these three aspects. Finally, we speculate on new trends in the development of this area.     

C-PSA: Constrained Pareto simulated annealing for constrained multi-objective optimization

In recent years, evolutionary algorithms (EAs) have been extensively developed and utilized to solve multi-objective optimization problems. However, some previous studies have shown that for certain problems, an approach which allows for non-greedy or uphill moves (unlike EAs), can be more beneficial. One such approach is simulated annealing (SA). SA is a proven heuristic for solving numerical optimization problems. But owing to its point-to-point nature of search, limited efforts has been made to explore its potential for solving multi-objective problems. The focus of the presented work is to develop a simulated annealing algorithm for constrained multi-objective problems. The performance of the proposed algorithm is reported on a number of difficult constrained benchmark problems. A comparison with other established multi-objective optimization algorithms, such as infeasibility driven evolutionary algorithm (IDEA), Non-dominated sorting genetic algorithm II (NSGA-II) and multi-objective Scatter search II (MOSS-II) has been included to highlight the benefits of the proposed approach.     

Evolving Multilayer Perceptrons

This paper proposes a new version of a method (G-Prop, genetic backpropagation) that attempts to solve the problem of finding appropriate initial weights and learning parameters for a single hidden layer Multilayer Perceptron (MLP) by combining an evolutionary algorithm (EA) and backpropagation (BP). The EA selects the MLP initial weights, the learning rate and changes the number of neurons in the hidden layer through the application of specific genetic operators, one of which is BP training. The EA works on the initial weights and structure of the MLP, which is then trained using QuickProp; thus G-Prop combines the advantages of the global search performed by the EA over the MLP parameter space and the local search of the BP algorithm. The application of the G-Prop algorithm to several real-world and benchmark problems shows that MLPs evolved using G-Prop are smaller and achieve a higher level of generalization than other perceptron training algorithms, such as QuickPropagation or RPROP, and other evolutive algorithms. It also shows some improvement over previous versions of the algorithm.     

Integrating Variable Reduction Strategy With Evolutionary Algorithms for Solving Nonlinear Equations Systems

Nonlinear equations systems (NESs) are widely used in real-world problems and they are difficult to solve due to their nonlinearity and multiple roots. Evolutionary algorithms (EAs) are one of the methods for solving NESs, given their global search capabilities and ability to locate multiple roots of a NES simultaneously within one run. Currently, the majority of research on using EAs to solve NESs focuses on transformation techniques and improving the performance of the used EAs. By contrast, problem domain knowledge of NESs is investigated in this study, where we propose the incorporation of a variable reduction strategy (VRS) into EAs to solve NESs. The VRS makes full use of the systems of expressing a NES and uses some variables (i.e., core variable) to represent other variables (i.e., reduced variables) through variable relationships that exist in the equation systems. It enables the reduction of partial variables and equations and shrinks the decision space, thereby reducing the complexity of the problem and improving the search efficiency of the EAs. To test the effectiveness of VRS in dealing with NESs, this paper mainly integrates the VRS into two existing state-of-the-art EA methods (i.e., MONES and DR-JADE) according to the integration framework of the VRS and EA, respectively. Experimental results show that, with the assistance of the VRS, the EA methods can produce better results than the original methods and other compared methods. Furthermore, extensive experiments regarding the influence of different reduction schemes and EAs substantiate that a better EA for solving a NES with more reduced variables tends to provide better performance.      

Inducing oblique decision trees with evolutionary algorithms

This paper illustrates the application of evolutionary algorithms (EAs) to the problem of oblique decision-tree (DT) induction. The objectives are to demonstrate that EAs can find classifiers whose accuracy is competitive with other oblique tree construction methods, and that, at least in some cases, this can be accomplished in a shorter time. We performed experiments with a (1+1) evolution strategy and a simple genetic algorithm on public domain and artificial data sets, and compared the results with three other oblique and one axis-parallel DT algorithms. The empirical results suggest that the EAs quickly find competitive classifiers, and that EAs scale up better than traditional methods to the dimensionality of the domain and the number of instances used in training. In addition, we show that the classification accuracy improves when the trees obtained with the EAs are combined in ensembles, and that sometimes it is possible to build the ensemble of evolutionary trees in less time than a single traditional oblique tree.     

Training and testing a self-adaptive multi-operator evolutionary algorithm for constrained optimization

Over the last two decades, many different evolutionary algorithms (EAs) have been introduced for solving constrained optimization problems (COPs). Due to the variability of the characteristics in different COPs, no single algorithm performs consistently over a range of practical problems. To design and refine an algorithm, numerous trial-and-error runs are often performed in order to choose a suitable search operator and the parameters. However, even by trial-and-error, one may not find an appropriate search operator and parameters. In this paper, we have applied the concept of training and testing with a self-adaptive multi-operator based evolutionary algorithm to find suitable parameters. The training and testing sets are decided based on the mathematical properties of 60 problems from two well-known specialized benchmark test sets. The experimental results provide interesting insights and a new way of choosing parameters.     

A review of evolutionary artificial neural networks

Research on potential interactions between connectionist learning systems, i.e., artificial neural networks (ANNs), and evolutionary search procedures, like genetic algorithms (GAs), has attracted a lot of attention recently. Evolutionary ANNs (EANNs) can be considered as the combination of ANNs and evolutionary search procedures. This article first distinguishes among three kinds of evolution in EANNs, i.e., the evolution of connection weights, of architectures, and of learning rules. Then it reviews each kind of evolution in detail and analyzes critical issues related to different evolutions. the review shows that although a lot of work has been done on the evolution of connection weights and architectures, few attempts have been made to understand the evolution of learning rules. Interactions among different evolutions are seldom mentioned in current research. However, the evolution of learning rules and its interactions with other kinds of evolution, play a vital role in EANNs. Finally, this article briefly describes a general framework for EANNs, which not only includes the aforementioned three kinds of evolution, but also considers interactions among them. © 1993 John Wiley & Sons, Inc.     

Polygene-based evolutionary algorithms with frequent pattern mining

In this paper, we introduce polygene-based evolution, a novel framework for evolutionary algorithms (EAs) that features distinctive operations in the evolutionary process. In traditional EAs, the primitive evolution unit is a gene, wherein genes are independent components during evolution. In polygene-based evolutionary algorithms (PGEAs), the evolution unit is a polygene, i.e., a set of co-regulated genes. Discovering and maintaining quality polygenes can play an effective role in evolving quality individuals. Polygenes generalize genes, and PGEAs generalize EAs. Implementing the PGEA framework involves three phases: (I) polygene discovery, (II) polygene planting, and (III) polygene-compatible evolution. For Phase I, we adopt an associative classification-based approach to discover quality polygenes. For Phase II, we perform probabilistic planting to maintain the diversity of individuals. For Phase III, we incorporate polygene-compatible crossover and mutation in producing the next generation of individuals. Extensive experiments on function optimization benchmarks in comparison with the conventional and state-of-the-art EAs demonstrate the potential of the approach in terms of accuracy and efficiency improvement.     

BP网络进化及其在雷达目标识别中的应用

针对常规ＢＰ神经网络的ＢＰ算法只能训练固定结构的神经网络,存在诸如易落入局部极值、没有引入提高泛化能力的训练机制等固有不足之处,以及一些神经网络进化算法的进化机制中存在的缺陷,本文提出一种ＢＰ神经网络进化算法,并用于高分辨雷达目标一维距离像的识别问题。实验结果表明,经所述方法优化后的神经网络结构简单、泛化能力优于ＢＰ算法和一些进化算法训练的网络。     

A Fast Clustering Based Evolutionary Algorithm for Super-Large-Scale Sparse Multi-Objective Optimization

During the last three decades,evolutionary algorithms(EAs) have shown superiority in solving complex optimization problems,especially those with multiple objectives and non-differentiable landscapes.However,due to the stochastic search strategies,the performance of most EAs deteriorates drastically when handling a large number of decision variables.To tackle the curse of dimensionality,this work proposes an efficient EA for solving super-large-scale multi-objective optimization problems with spa...     

A space search optimization algorithm with accelerated convergence strategies

Evolutionary algorithms (EAs), which have been widely used to solve various scientific and engineering optimization problems, are essentially stochastic search algorithms operating in the overall solution space. However, such random search mechanism may lead to some disadvantages such as a long computing time and premature convergence. In this study, we propose a space search optimization algorithm (SSOA) with accelerated convergence strategies to alleviate the drawbacks of the purely random search mechanism. The overall framework of the SSOA involves three main search mechanisms: local space search, global space search, and opposition-based search. The local space search that aims to form new solutions approaching the local optimum is realized based on the concept of augmented simplex method, which exhibits significant search abilities realized in some local space. The global space search is completed by Cauchy searching, where the approach itself is based on the Cauchy mutation. This operation can help the method avoid of being trapped in local optima and in this way alleviate premature convergence. An opposition-based search is exploited to accelerate the convergence of space search. This operator can effectively reduce a substantial computational overhead encountered in evolutionary algorithms (EAs). With the use of them SSOA realizes an effective search process. To evaluate the performance of the method, the proposed SSOA is contrasted with a method of differential evolution (DE), which is a well-known space concept-based evolutionary algorithm. When tested against benchmark functions, the SSOA exhibits a competitive performance vis-a-vis performance of some other competitive schemes of differential evolution in terms of accuracy and speed of convergence, especially in case of high-dimensional continuous optimization problems.     

A review of genetic algorithms applied to training radial basis function networks

The problems associated with training feedforward artificial neural networks (ANNs) such as the multilayer perceptron (MLP) network and radial basis function (RBF) network have been well documented. The solutions to these problems have inspired a considerable amount of research, one particular area being the application of evolutionary search algorithms such as the genetic algorithm (GA). To date, the vast majority of GA solutions have been aimed at the MLP network. This paper begins with a brief overview of feedforward ANNs and GAs followed by a review of the current state of research in applying evolutionary techniques to training RBF networks.     

Rotated neighbor learning-based auto-configured evolutionary algorithm基于旋转邻域学习的自配置进化算法

More and more evolutionary operators have been integrated and manually configured together to solve wider range of problems. Considering the very limited progress made on the automatic configuration of evolutionary algorithms (EAs), a rotated neighbor learning-based auto-configured evolutionary algorithm (RNLACEA) is presented. In this framework, multiple EAs are combined as candidates and automatically screened for different scenarios with a rotated neighbor structure. According to a ranking record and a group of constraints, the algorithms can be better scheduled to improve the searching efficiency and accelerate the searching pace. Experimental studies based on 14 classical EAs and 22 typical benchmark problems demonstrate that RNLACEA outperforms other six representative auto-adaptive EAs and has high scalability and robustness in solving different kinds of numerical optimization problems.     

Using evolutionary algorithms as instance selection for data reduction in KDD: an experimental study

Evolutionary algorithms are adaptive methods based on natural evolution that may be used for search and optimization. As data reduction in knowledge discovery in databases (KDDs) can be viewed as a search problem, it could be solved using evolutionary algorithms (EAs). In this paper, we have carried out an empirical study of the performance of four representative EA models in which we have taken into account two different instance selection perspectives, the prototype selection and the training set selection for data reduction in KDD. This paper includes a comparison between these algorithms and other nonevolutionary instance selection algorithms. The results show that the evolutionary instance selection algorithms consistently outperform the nonevolutionary ones, the main advantages being: better instance reduction rates, higher classification accuracy, and models that are easier to interpret.     

进化算法研究进展

进化算法是一类借鉴生物界自然选择和自然遗传机制的随机搜索算法，主要包括遗传算法，（ｇｅｎｅｒｉｃａｌｇｏｒｉｔｈｍｓ，简记为ＧＡｓ）、进化规划（ｅｖｏｌｕｔｉｏｎａｒｙｐｒｏｇｒａｍｍｉｎｇ，简记为ＥＰ）和进化策略（ｅｖｏｌｕｔｉｏｎａｒｙｓｔｒａｔｅｇｉｅｓ，简记为ＥＳｓ），它们可以用解决优化和机器学习等问题，进化算法的两个主要特点中群体搜索策略及群体中个体之间的信息交换，进化算法不依赖于梯度信     

Target shape design optimization by evolving B-splines with cooperative coevolution

With high reputation in handling non-linear and multi-model problems with little prior knowledge, evolutionary algorithms (EAs) have successfully been applied to design optimization problems as robust optimizers. Since real-world design optimization is often computationally expensive, target shape design optimization problems (TSDOPs) have been frequently used as efficient miniature model to check algorithmic performance for general shape design. There are at least three important issues in developing EAs for TSDOPs, i.e., design representation, fitness evaluation and evolution paradigm. Existing work has mainly focused on the first two issues, in which (1) an adaptive encoding scheme with B-spline has been proposed as a representation, and (2) a symmetric Hausdorff distance based metric has been used as a fitness function. But for the third issue, off-the-shelf EAs were used directly to evolve B-spline control points and/or knot vector. In this paper, we first demonstrate why it is unreasonable to evolve the control points and knot vector simultaneously. And then a new coevolutionary paradigm is proposed to evolve the control points and knot vector of B-spline separately in a cooperative manner. In the new paradigm, an initial population is generated for both the control points, and the knot vector. The two populations are evolved mostly separately in a round-robin fashion, with only cooperation at the fitness evaluation phase. The new paradigm has at least two significant advantages over conventional EAs. Firstly, it provides a platform to evolve both the control points and knot vector reasonably. Secondly, it reduces the difficulty of TSDOPs by decomposing the objective vector into two smaller subcomponents (i.e., control points and knot vector). To evaluate the efficacy of the proposed coevolutionary paradigm, an algorithm named CMA-ES-CC was formulated. Experimental studies were conducted based on two target shapes. The comparison with six other EAs suggests that the proposed cooperative coevolution paradigm is very effective for TSDOPs.     

约束优化进化算法

约束优化问题是科学和工程应用领域经常会遇到的一类数学规划问题.近年来,约束优化问题求解已成为进化计算研究的一个重要方向.从约束优化进化算法=约束处理技术+进化算法的研究框架出发,从约束处理技术和进化算法两个基本方面对约束优化进化算法的研究及进展进行了综述.此外,对约束优化进化算法中的一些重要问题进行了探讨.最后进行了各种算法的比较性总结,深入分析了目前约束优化进化算法中亟待解决的问题,并指出了值得进一步研究的方向.