数据驱动选择策略的多目标差分进化算法

针对多目标差分进化算法求解复杂多目标优化问题时,最优解选择策略中非支配排序计算复杂度高的问题,提出一种数据驱动选择策略的多目标差分进化(MODE-DDSS)算法.首先,设计多目标差分进化算法的优化解排序等级评估准则,建立基于评估准则的优化解排序等级评估库;其次,设计基于优化解双向搜索机制和无重复比较机制的数据驱动选择策略,实现优化解的高效搜索和快速排序;最后,构建数据驱动选择策略的多目标差分进化算法,降低算法在最优解选择操作中的时间复杂度,提高算法的寻优效率.实验结果表明,所提出的MODE-DDSS算法能够有效减少最优解在选择过程中的比较次数,提升多目标差分进化算法解决复杂多目标优化问题的寻优效率.     

基于个体适应度梯度的定向进化算法

基于进化过程中父代个体和子代个体在种群中的适应度梯度,提出进化方向的概念,并对其进行定性分析,在此基础上提出最优进化方向。通过最优进化方向的指导可加速进化过程,提高进化算法的收敛效率。基于进化方向和最优进化方向的描述,设计并实现基于个体适应度梯度的定向进化算法,并针对该算法给出2种不同的个体繁殖策略。对算法的收敛性和复杂度进行理论分析。最后通过仿真实验说明该算法具有精度高、收敛速度快等优点,在一定程度上克服目前进化算法的搜索低效问题。     

一种求解作业车间调度的混合粒子群算法*

针对车间作业调度问题,提出了一种混合了知识进化算法和粒子群优化的算法。算法主要是结合知识进化算法的进化选择机制和粒子群优化的局部快速收敛性特性,首先让粒子替代知识进化算法中的进化个体,在群体空间中按粒子群优化规则寻找局部最优,然后根据知识进化算法的全局选择机制寻找全局最优,最后,将车间作业调度问题的特点融入到所提出的混合算法中求解问题。采用基准数据进行测试的仿真实验,并比对标准遗传算法,结果表明所提算法的有效性。     

一种进化半监督式模糊聚类的入侵检测算法

在入侵检测系统中,未知标签数据容易获得,标签数据较难获得,对此提出了一种基于进化半监督式模糊聚类入侵检测算法。算法利用标签数据信息担任染色体的角色,引导非标签数据每个模糊分类的进化过程,能够使用少量的标签数据和大量未知标签数据生成入侵检测系统分类器,可处理模糊类标签,不易陷入局部最优,适合并行结构的实现。实验结果表明,算法有较高的检测率。     

基于动态双子种群的差分进化K中心点聚类算法

随着海量大数据的出现,聚类算法需要新型计算模式来提高计算速度与运行效率。本文提出一种基于动态双子种群的差分进化K中心点聚类算法DGP-DE-K-mediods(Dynamic Gemini Population based DE-K-mediods)。DGP-DE-K-mediods利用动态双子种群方法,解决聚类算法在维持种群密度的时候避免陷入局部最优的问题;采用差分进化(Differential Evolution, DE)算法来提高全局最优能力的强健性;基于Hadoop云平台来并行处理DGP-DE-K-mediods,加快算法的运行速度和效率;描述基于MapReduce的并行聚类算法的编程过程;DGP-DE-K-mediods利用UIC的大数据分类的案例数据和网络入侵检测这种大数据应用来仿真算法的效果。实验结果表明,与已有的聚类算法相比,DGP-DE-K-mediods在检测精度、运行时间上有明显的优势。     

基于量子免疫算法的车辆调度问题优化

为优化带时间窗的车辆调度计算问题,引入量子进化算法,提出了一种混合量子免疫进化算法。首先对传统量子旋转门进行改进,使个体在进化过程中向全局最优位置靠近,从而避免算法早熟并保持种群多样性。其次在迭代过程中,引入免疫算子,提取优秀基因片段作为疫苗,接种到种群中其他个体,避免算法性能的倒退。最后,针对Solomon标准实例库实例数据进行多算法编码仿真实验,结果表明,所提混合量子免疫进化算法不仅能够有效解决类似问题,而且能够显著加速收敛。     

融合自控粒子群和免疫进化的入侵数据分类

在对基于异常的入侵检测进行训练时,缺少一个实时有效的训练集,提出了一种融合自控粒子群和免疫进化的入侵数据分类方法,对网络数据进行聚类分析,生成可靠的训练数据。粒子群模糊C均值聚类算法需要提前确定聚类数目,这在网络数据分析处理中是很难把握的,引入自控粒子群的方法根据迭代演算情况自动调节不同聚类数目的粒子群规模,使数据最后聚合在一个数目最优的聚类集中,同时为了克服陷入局部最优的问题,引入免疫进化机制,使部分粒子在当前最优指导下进行合理变异和替换,跳出局部最优解。     

求解多目标作业车间调度问题的混合变异杂草优化算法

针对多目标作业车间调度问题,提出一种混合变异杂草优化算法。该算法采用基于各子目标熵值权重的欧氏贴近度作为适应度值计算方法,引导种群向Pareto前端进化。在进化过程中,运用快速非支配排序策略构建Pareto档案,并利用进化种群中最优个体实时更新Pareto最优解集,提升算法的优化性能;同时通过引入变异算子增加种群多样性,避免算法陷入局部最优。最后,基于Benchmark算例的仿真实验,验证了该算法求解多目标作业车间调度问题的有效性。     

一种基于Pareto最优概念和神经网络混合策略的多目标进化算法

基于Pareto最优的多目标进化算法得到了广泛地应用,但不适用于目标函数为非解析式的情况。基于神经网络和Pareto最优的联合策略,提出了一种解决此类问题的新方法：首先采用神经网络对历史数据进行学习,建立有效的神经网络模型来代替目标函数解析式;然后将神经网络模型嵌入多目标进化算法,进行进化计算;最后,将本文方法应用于卷烟配方比例掺配问题。实验结果表明,该方法优于传统方法,能较好地解决问题。     

全局优化径向基函数网络方法的设计及应用

针对径向基神经网络（RBFNN）中隐层单元中心及输出层权值向量难以有效确定的问题,论文提出了一种基于进化思想的解决方案.以进化算法中的模式定理为理论依据,运用分治策略思想,将隐层中心的最优化过程和输出层权值向量的最优化过程并行处理,提高了算法效率.最后将设计的整体进化径向基神经网络应用于数据的分类,以UCI数据库中的iris和wheat数据集为测试物料,采用该文提出的进化方案得到最优中心和权值向量,测试表明相对于RBFNN运用聚类办法确定中心和最小二乘确定权值的方法以及支持向量机其检出率能提高20％.     

基于单亲遗传算法的RoboCup动态角色分配

RoboCup的机器人动态角色分配问题是一个典型的组合优化问题。解决这一问题的传统方法是贪心法,但贪心法易陷入局部最优解。提出用针对组合优化问题而构造的序号编码单亲遗传算法解决RoboCup的机器人动态角色分配问题。单亲遗传算法借鉴了传统遗传算法“优胜劣汰”的自然选择机制,但只通过单个体繁殖后代,在解决组合优化问题和复杂工程优化问题方面具有明显的优越性。试验结果显示这种方法的在解决RoboCup机器人动态角色分配问题时的有效性。     

Evolutionary algorithms refining a heuristic: a hybrid method for shared-path protections in WDM networks under SRLG constraints.

An evolutionary algorithm (EA) can be used to tune the control parameters of a construction heuristic to an optimization problem and generate a nearly optimal solution. This approach is in the spirit of indirect encoding EAs. Its performance relies on both the heuristic and the EA. This paper proposes a three-phase parameterized construction heuristic for the shared-path protection problem in wavelength division multiplexing networks with shared-risk link group constraints and applies an EA for optimizing the control parameters of the proposed heuristics. The experimental results show that the proposed approach is effective on all the tested network instances. It was also demonstrated that an EA with guided mutation performs better than a conventional genetic algorithm for tuning the control parameters, which indicates that a combination of global statistical information extracted from the previous search and location information of the best solutions found so far could improve the performance of an algorithm.     

A real multi-parent tri-hybrid evolutionary optimization method and its application in wind velocity estimation from wind profiler data

A real-coded multi-parent tri-hybrid evolutionary algorithm (EA) for problem optimization is presented. The hybrid EA algorithm combines the features of Simplex, stochastic relaxation (SR) and multi-parent EA reproduction in a model that encourages competition among the best individual solutions from various operations. Its strength has been evaluated using standard test functions and shown to do better than other methods. The algorithm’s ability to handle noise is evident when applied to experiments involving resolution of overlapping wind profiler (WP) data. Results obtained using raw data closely matched those obtained with data preprocessed by a low-pass FFT filter. Resolution of low-speed wind and clutter signals in various degrees of overlap is made possible, thereby allowing the determination of wind velocity and variance to be executed with ease.     

射频供能传感网面向融合检测的部署调度方法

当射频供能传感网应用于目标检测时,对节点的部署位置和充电/感知调度表进行合理规划可有效提高系统检测质量.基于融合检测模型,首先归纳了使得系统检测质量最大化的节点部署和调度联合优化问题,证明了该问题是NP完全问题.然后分析了融合半径对检测率的影响,提出了基于贪婪算法的节点部署调度联合优化算法.通过小规模网络、大规模网络及基于真实数据集的仿真,将该算法分别与全局最优解、分阶段优化贪婪算法进行了性能比较.实现结果表明：所提出的联合优化贪婪算法获得的系统检测质量在各组仿真中均优于分阶段贪婪算法,并且在小规模网络中接近于全局最优解.     

A hybridization of an evolutionary algorithm and a parallel branch and bound for solving the capacitated single allocation hub location problem

In this study, we propose a hybrid optimization method, consisting of an evolutionary algorithm (EA) and a branch-and-bound method (BnB) for solving the capacitated single allocation hub location problem (CSAHLP). The EA is designed to explore the solution space and to select promising configurations of hubs (the location part of the problem). Hub configurations produced by the EA are further passed to the BnB search, which works with fixed hubs and allocates the non-hub nodes to located hubs (the allocation part of the problem). The BnB method is implemented using parallelization techniques, which results in short running times. The proposed hybrid algorithm, named EA-BnB, has been tested on the standard Australia Post (AP) hub data sets with up to 300 nodes. The results demonstrate the superiority of our hybrid approach over existing heuristic approaches from the existing literature. The EA-BnB method has reached all the known optimal solutions for AP hub data set and found new, significantly better, solutions on three AP instances with 100 and 200 nodes. Furthermore, the extreme efficiency of the implementation of this hybrid algorithm resulted in short running times, even for the largest AP test instances.     

Evolutionary optimization of radial basis function classifiers for data mining applications.

In many data mining applications that address classification problems, feature and model selection are considered as key tasks. That is, appropriate input features of the classifier must be selected from a given (and often large) set of possible features and structure parameters of the classifier must be adapted with respect to these features and a given data set. This paper describes an evolutionary algorithm (EA) that performs feature and model selection simultaneously for radial basis function (RBF) classifiers. In order to reduce the optimization effort, various techniques are integrated that accelerate and improve the EA significantly: hybrid training of RBF networks, lazy evaluation, consideration of soft constraints by means of penalty terms, and temperature-based adaptive control of the EA. The feasibility and the benefits of the approach are demonstrated by means of four data mining problems: intrusion detection in computer networks, biometric signature verification, customer acquisition with direct marketing methods, and optimization of chemical production processes. It is shown that, compared to earlier EA-based RBF optimization techniques, the runtime is reduced by up to 99% while error rates are lowered by up to 86%, depending on the application. The algorithm is independent of specific applications so that many ideas and solutions can be transferred to other classifier paradigms.     

基于并行遗传算法的配送路线求解

遗传算法(Genetic Algorithm)是一类借鉴生物界的进化规律演化而来的随机化搜索方法,已经成功运用在很多大规模的组合优化问题中。利用如今流行的并行计算机系统,对遗传算法进行并行化,可解决标准遗传算法的速度瓶颈问题。本文在MPI并行环境下,用C++语言实现了粗粒度模型的并行遗传算法。结合并行遗传算法的特点,提出了解决物流配送路线优化的策略以及给出相应的算法过程,并进行了有效验证。通过研究结果表明,与传统遗传算法相比,并行遗传算法提高了运算速度,降低了平均开销时间并且最小总路径值更理想。     

An Evolutionary Algorithm for Global Optimization Based on Level-Set Evolution and Latin Squares

In this paper, the level-set evolution is exploited in the design of a novel evolutionary algorithm (EA) for global optimization. An application of Latin squares leads to a new and effective crossover operator. This crossover operator can generate a set of uniformly scattered offspring around their parents, has the ability to search locally, and can explore the search space efficiently. To compute a globally optimal solution, the level set of the objective function is successively evolved by crossover and mutation operators so that it gradually approaches the globally optimal solution set. As a result, the level set can be efficiently improved. Based on these skills, a new EA is developed to solve a global optimization problem by successively evolving the level set of the objective function such that it becomes smaller and smaller until all of its points are optimal solutions. Furthermore, we can prove that the proposed algorithm converges to a global optimizer with probability one. Numerical simulations are conducted for 20 standard test functions. The performance of the proposed algorithm is compared with that of eight EAs that have been published recently and the Monte Carlo implementation of the mean-value-level-set method. The results indicate that the proposed algorithm is effective and efficient.     

A Constraint-Based Evolutionary Learning Approach to the Expectation Maximization for Optimal Estimation of the Hidden Markov Model for Speech Signal Modeling

This paper attempts to overcome the tendency of the expectation-maximization (EM) algorithm to locate a local rather than global maximum when applied to estimate the hidden Markov model (HMM) parameters in speech signal modeling. We propose a hybrid algorithm for estimation of the HMM in automatic speech recognition (ASR) using a constraint-based evolutionary algorithm (EA) and EM, the CEL-EM. The novelty of our hybrid algorithm (CEL-EM) is that it is applicable for estimation of the constraint-based models with many constraints and large numbers of parameters (which use EM) like HMM. Two constraint-based versions of the CEL-EM with different fusion strategies have been proposed using a constraint-based EA and the EM for better estimation of HMM in ASR. The first one uses a traditional constraint-handling mechanism of EA. The other version transforms a constrained optimization problem into an unconstrained problem using Lagrange multipliers. Fusion strategies for the CEL-EM use a staged-fusion approach where EM has been plugged with the EA periodically after the execution of EA for a specific period of time to maintain the global sampling capabilities of EA in the hybrid algorithm. A variable initialization approach (VIA) has been proposed using a variable segmentation to provide a better initialization for EA in the CEL-EM. Experimental results on the TIMIT speech corpus show that CEL-EM obtains higher recognition accuracies than the traditional EM algorithm as well as a top-standard EM (VIA-EM, constructed by applying the VIA to EM).     

Virulence Optimization Algorithm

In this paper, a new optimization algorithm to solve continuous and non-linear optimization problems is introduced. This algorithm is inspired by the optimal mechanism of viruses when infecting body cells. Special mechanism and function of viruses which includes the recognition of fittest viruses to infect body cells, reproduction (cloning) of these cells to prompt “invasion” operation of ready-to-infect regions and then escaping from infected regions (to avoid immune reaction) is the basis of this evolutionary optimization algorithm. Like many evolutionary algorithms, the Virulence Optimization Algorithm (VOA) starts the optimization process with an initial population consisting of viruses and host cells. The host cell population represents the resources available in host environment or the region containing the global optimum solution. The virus population infiltrates the host environment and attempts to infect it. In the optimization procedure, at first the viruses reside in the constituted regions or clusters of the environment called virus groups (via K-means clustering). Then they scatter in host environment through mutation (Drifting) and recombination (Shifting) operators. Then among the virus groups, the group with highest mean fitness is chosen as escape destination. Before the escape operation commences, the best viruses in each virus group are recognized and undergoes a cloning operation to spread the Virulence in the host environment. This procedure continues until the majority of the virus population is gathered in the region containing the maximum resources or the global optimum solution. The novelty of the proposed algorithm is achieved by simulating three important and major mechanisms in the virus life, namely (1) the reproduction and mutation mechanism, (2) the cloning mechanism to generate the best viruses for rapid and excessive infection of the host environment and (3) the mechanism of escaping from the infected region. Simulating the first mechanism in the virus life enables the proposed algorithm to generate new and fittest virus varieties. The cloning mechanism facilitates the excessive spread of the fittest viruses in the host environment to infect the host environment more quickly. Also, to avoid the immune response, the fittest viruses (with a great chance of survival) are duplicated through the cloning process, and scattered according to the Vicinity Region Radius of each region. Then, the fittest viruses escape the infected region to reside in a region which possess the resources necessary to survive (global optimum). The evaluation of this algorithm on 11 benchmark test functions has proven its capability to deal with complex and difficult optimization problems.