基于种群多样性控制的多级信息迁移多任务优化粒子群算法

基于群体智能“隐并行性”实现多任务优化已取得一系列研究成果,但任务间频繁的垂直信息传递导致种群异质性过度增加,进而产生信息负迁移消极影响,这也是目前多任务优化领域尚未完全解决的难题之一.针对此问题,首先将粒子群算法(PSO)与多种群演化信息共享机制相结合,然后引入标杆管理思想实现多层级信息迁移及智能涌现,最后通过计算种群多样性指数有效控制信息迁移频率,提出多级信息迁移多任务优化PSO算法(multi-level information transfer multi-task PSO, MITMPSO).仿真实验表明,通过设置合理的信息迁移阈值,MITMPSO能在多项式时间内显著提高多任务高维函数优化、多任务多约束函数优化以及多任务二元离散优化问题的求解质量,加快各优化问题的收敛速度.     

多任务处理协同进化粒子群算法

粒子群算法及其改进算法专注于单任务的求解.随着电子商务的发展,在线服务器在某一时刻需要同时处理来自多个用户的服务请求,即多任务处理.区别于以往的并行计算机,文中充分挖掘粒子群算法的“隐并行性”,并引入协同进化机制,在同一搜索空间根据任务个数设置不同的子种群,各子种群以一定的概率相互传递有效信息,最后提出基于多任务处理协同进化粒子群算法(CPSOM),并将CPSOM应用于多任务连续型函数优化问题、多任务离散型属性选择问题以及多任务约束工程优化问题.仿真实验表明,在CPSOM多任务环境中,不同任务之间确实存在有效信息的传递,不同任务之间的相互协作不仅可以提高解的质量,而且可以加快各优化问题的收敛速度.     

面向路径规划问题的虚拟多任务共生生物搜索算法

针对现有共生生物搜索(SOS)算法在求解路径规划等离散型优化问题时存在性能较差、收敛速度慢等问题，提出虚拟多任务共生生物搜索(VMTSOS)算法。首先根据双向映射解码策略，实现个体连续空间位置和离散城市序列转换；然后引入多任务优化思想构建虚拟多任务环境，设计多种群同时优化同一任务，并通过停滞阈值控制种群间信息迁移频率，当主种群达到停滞阈值时，将辅助种群中部分优秀个体替换为主种群劣质个体；最后对VMTSOS算法时间和空间复杂度进行分析。仿真实验表明，VMTSOS算法在求解多数TSP时均能快速收敛至各测试实例目前的最优解，而在求解冷链物流配送问题时，具有多种群辅助机制的VMTSOS算法能较大程度地降低最优总成本。     

多任务优化算法综述

基于计算智能“隐并行性”实现多任务优化(multi-task optimization, MTO),是当前研究的热点和前沿技术.与传统单任务优化算法相比,通过挖掘群体智能内在并行和内涵并行同时优化多个任务,可显著提高问题求解质量以及缩短任务求解时间.首先,对MTO相关英文/中文文献进行梳理,总结MTO研究进展和趋势;然后,基于多因子优化(multifactorial optimization, MFO)和多种群演化(multi-population evolution, MPE)两种不同信息共享框架,从多任务搜索空间设计、种群数量、种群规模、依托算法、信息迁移节点、交互信息、时间和空间复杂度以及复杂系统等角度对比二者异同;接着,从信息迁移节点、方式和类型3方面重点阐述MTO核心理论;最后,从探究MTO复杂系统层级智能涌现行为、多任务种群多样性控制以及应用领域拓展3方面展望未来研究方向.     

信息筛选多任务优化自组织迁移算法

针对现有自组织迁移算法（SOMA）只能求解单个优化问题及其“隐并行性”未能被充分挖掘的缺陷,提出信息筛选多任务优化自组织迁移算法（SOMAMIF）实现同一时刻处理多个优化问题。首先,构造多任务统一搜索空间,并根据任务个数设置相应的子种群;然后,对各子种群当前最优适应值进行判断,当任务连续若干代停滞进化时则产生信息交互需求;接着,按概率从剩余子种群中筛选对自己有用的信息并过滤无用信息,从而在保证信息正向迁移同时实现种群结构的重新调整;最后对算法的时间复杂度和空间复杂度进行分析。实验结果表明,SOMAMIF在同时求解多个高维函数优化问题时均快速收敛至全局最优解0,而SOMAMIF与分形技术相结合同时提取不同户籍高校学生返乡关键制约因素时,其在两个数据集上得到的平均分类准确率与原始数据集的平均分类准确率相比分别提高了0.348 66个百分点和0.598 57个百分点。     

信息迁移多任务优化共生生物搜索算法

针对现有共生生物搜索（SOS）算法只能求解单个任务，以及信息负迁移影响多任务优化（MTO）性能这两个难题，提出一个信息迁移多任务优化共生生物搜索（ITMTSOS）算法。首先基于多种群演化MTO框架，根据任务个数设置相应数量种群；然后各种群独立运行基本SOS算法，当某一种群连续若干代停滞进化时，引入个体自身最优经验和邻域最优个体以形成知识模块并将该模块迁移至该种群个体进化过程中；最后对ITMTSOS算法时间和空间复杂度进行分析。仿真实验结果表明，ITMTSOS算法同时求解多个不同形态高维函数时均能快速收敛至全局极值解0，与单任务SOS算法相比，平均运行时间最多缩短约25.25%；而在同时求解多维0/1背包问题和师生匹配问题时，所提算法在测试集weing1和weing7上的最优适应值与目前测试集公布的最优结果相比分别提高了22 767和22 602，师生最优匹配差和平均匹配差的绝对值分别下降了26和33，平均运行时间约缩短了7.69%。     

基于Q学习的多任务多目标粒子群优化算法

多任务粒子群优化算法(multi-task particle swarm ptimization, MTPSO)通过知识迁移学习,具有快速收敛能力,广泛应用于求解多任务多目标优化问题.然而, MTPSO难以根据种群进化状态自适应调整优化过程,容易陷入局部最优,收敛性能较差.针对此问题,利用强化学习的自我进化与预测能力,提出一种基于Q学习的多任务多目标粒子群优化算法(QM2PSO).首先,设计粒子群参数动态更新方法,利用Q学习方法在线更新粒子群算法的惯性权重和加速度参数,提高当前粒子收敛到Pareto前沿的能力;其次,提出基于柯西分布的突变搜索策略,通过全局和局部交替搜索多任务最优解,避免算法陷入局部最优;最后,设计基于正向迁移准则的知识迁移方法,采用Q学习方法更新知识迁移率,改善知识负迁移现象.与已有经典算法的对比实验结果表明所提出的QM2PSO算法具有更优越的收敛性.     

教与学信息交互粒子群优化算法

针对单一种群在解决高维问题中收敛速度较慢和多样性缺失的问题,提出了一种教与学信息交互粒子群优化（PSO）算法。根据进化过程将种群动态地划分为两个子种群,分别采用粒子群优化算法和教与学优化算法,同时粒子利用学习者阶段进行子种群之间信息交互,并通过评价收敛性和多样性指标让粒子的收敛能力和多样性在进化过程中得到平衡。与粒子群优化算法、混合灰狼粒子群算法、重选精英个体的非线性收敛灰狼优化（GWO）算法等多个进化算法在15个标准测试函数的不同维度下进行对比实验,所提算法在多个测试函数上可以收敛到理论最优值,速度相对于其他算法提高了1~6倍。实验结果表明,所提算法在收敛精度和收敛速度上具有较好的效果。     

基于机器学习的演化多任务优化框架

演化多任务优化是近年来计算智能领域的研究热点之一,其原理是通过任务间的知识转移提高演化算法同时求解多个任务的效率.由于任务间相似性对促进任务之间的正向知识转移具有重要的影响,因此,如何度量任务间的相似性成为了重点研究方向之一.目前,演化多任务优化在处理两个任务时,辅助任务的选取仅限于两者之一,且在处理超多任务时对任务间知识的转移缺乏灵活性.为此,本文提出一个基于机器学习的演化多任务优化框架,命名为MaTML.该框架联合所有任务关联的子种群形成一个统一的初始化种群,利用目标任务的技能因子及其对应的种群个体分别构建标签和训练集,应用十折交叉法拟合模型,并运用模型预测与目标任务相似的个体以组成辅助种群,从而促进演化优化中的正向知识转移.本文提出的算法能够在动态的种群个体中找到目标任务的辅助种群,不仅可以为三个或以上的多任务优化灵活地选取相似辅助任务,而且解决了当任务数量为两个时有效地选择辅助任务的问题.通过与现阶段的多任务算法和超多任务算法分别在CEC2017问题测试集和WCCI2020SO问题测试集进行比较,实验结果证实MaTML在优化多任务问题时具有更优或竞争性的性能.此外,文中还详细研...     

求解测试用例自动生成问题的多因子回溯搜索优化算法

路径覆盖测试用例自动生成（ATCG-PC）问题是自动化软件测试领域的热点。ATCG-PC问题中群智能进化算法常用的适应度函数之间具有高度的相似性，然而现有的解决ATCG-PC问题的群智能进化算法尚未考虑这一相似性特征。受相似性特征启发，将两个相似的适应度函数看作两个任务，从而将ATCG-PC问题转化为多任务ATCGPC问题，并提出了一种新的解决多任务ATCG-PC问题的群智能进化算法，即多因子回溯搜索优化算法（MFBSA）。所提算法通过多因子选择Ⅰ的记忆种群功能提高全局搜索能力，并通过选型记忆交配使得相似任务之间能够通过知识转移提高彼此的优化效率。在6个雾计算测试程序和6个自然语言处理测试程序上对所提算法性能进行了评价。与回溯搜索优化算法（BSA）、免疫遗传算法（IGA）、收敛速度控制器粒子群优化（PSO-CSC）算法、自适应粒子群优化（APSO）算法和超立方体差分进化（DE-H）算法相比，MFBSA覆盖12个测试程序上的路径所使用的测试用例总数分别减少了64.46%、66.64%、67.99%、74.15%和61.97%。实验结果表明，所提算法能够有效降低测试成本。     

动态评价免疫微粒群算法在Job-shop调度中的应用

传统粒子群优化算法在解决组合优化问题上具有一定的局限性,通过分析其优化机理,对迭代公式加以改进,提出了改进微粒群算法。算法中,利用遗传算法的交叉思想来完成粒子间的信息交换,以期达到粒子更新。粒子进化过程中,为保留群体中的优秀粒子,使用了加速度这一优化算子。为避免粒子陷入局部搜索,迭代过程中使用免疫算法来动态评价微粒群体。通过大量实验仿真,算法可以有效求解作业车间调度问题,验证了算法的合理性。     

Why do media users multitask?: Motives for general,medium-specific,and content-specific types of multitasking

This study examined the major motives for multitasking, and how those motives are related to general, medium-specific, and content-specific types of multitasking. The major motives for multitasking identified in this study are as follows: information, social, enjoyment, efficiency, and habit. Of these motives, general multitasking behavior was predicted by information, efficiency, and habit. In terms of medium-specific types of multitasking, TV-based multitasking was predicted by habit motive, Internet-based multitasking was predicted by information and enjoyment, and mobile-based multitasking was predicted by information motives. In terms of content-specific multitasking, news-related multitasking was predicted by information motives, entertainment-related multitasking was predicted by information and enjoyment motives, and advertising-related multitasking was predicted by information and social motives.     

基于混沌搜索的粒子群优化算法

粒子群优化算法(PSO)是一种有效的随机全局优化技术。文章把混沌优化搜索技术引入到PSO算法中,提出了基于混沌搜索的粒子群优化算法。该算法保持了PSO算法结构简单的特点,改善了PSO算法的全局寻优能力,提高的算法的收敛速度和计算精度。仿真计算表明,该算法的性能优于基本PSO算法。     

Cooperative particle swarm optimizer with depth first search strategy for global optimization of multimodal functions

This paper presents a Cooperative Particle Swarm Optimizer with Depth First Search Strategy (DFS-CPSO), which has better seacrch capality than classical Particle Swarm Optimizer (PSO) in solving multimodal optimization problems. In order to improve the quality of information exchange, the Depth First Search (DFS) strategy is hybridized to Cooperative Particle Swarm Optimization(CPSO), which makes information transfer more effectively and generates better quality solution. Specifically, DFS strategy enables different components of solution vector to exchange information separately with PSO and increases the diversity of the population, so that the information of solution components could be preserved by multiple iterations in CPSO. Confirmatory experiments are performed to prove the effectiveness of employing the DFS strategy to CPSO. The comparative results demonstrate superior performance of DFS-CPSO in solving high dimensional multimodal functions than CPSO and other advanced methods.

     

In-class multitasking and academic performance

The omnipresence of student-owned information and communication technologies (ICTs) in today’s college classrooms presents educational opportunities but can also create learning problems. Specifically, multitasking with these technologies can interfere with the learning process. Indeed, research in cognitive science shows that there are clear performance decrements when trying to attend to two tasks at the same time. This study examines the frequency with which students multitask during class using a large sample (N = 1,839) and examines the relationship between multitasking and academic performance as measured by actual overall semester grade point average (GPA). Students reported frequently text messaging during class but reported multitasking with other ICTs to a lesser extent. Furthermore, only social technologies (Facebook and text messaging) were negatively related to GPA.     

Recent approaches to global optimization problems through Particle Swarm Optimization

This paper presents an overview of our most recent results concerning the Particle Swarm Optimization (PSO) method. Techniques for the alleviation of local minima, and for detecting multiple minimizers are described. Moreover, results on the ability of the PSO in tackling Multiobjective, Minimax, Integer Programming and ¹ errors-in-variables problems, as well as problems in noisy and continuously changing environments, are reported. Finally, a Composite PSO, in which the heuristic parameters of PSO are controlled by a Differential Evolution algorithm during the optimization, is described, and results for many well-known and widely used test functions are given.     

Non-parametric particle swarm optimization for global optimization

In recent years, particle swarm optimization (PSO) has extensively applied in various optimization problems because of its simple structure. Although the PSO may find local optima or exhibit slow convergence speed when solving complex multimodal problems. Also, the algorithm requires setting several parameters, and tuning the parameters is a challenging for some optimization problems. To address these issues, an improved PSO scheme is proposed in this study. The algorithm, called non-parametric particle swarm optimization (NP-PSO) enhances the global exploration and the local exploitation in PSO without tuning any algorithmic parameter. NP-PSO combines local and global topologies with two quadratic interpolation operations to increase the search ability. Nineteen (19) unimodal and multimodal nonlinear benchmark functions are selected to compare the performance of NP-PSO with several well-known PSO algorithms. The experimental results showed that the proposed method considerably enhances the efficiency of PSO algorithm in terms of solution accuracy, convergence speed, global optimality, and algorithm reliability.     

Why Some People Multitask Better Than Others: Predicting Learning

Multitasking using information systems (IS) often results in a blurring between personal and work lives. Prior literature overwhelmingly points to negative effects of traditional multitasking. Surprisingly, research on IS multitasking is relatively scarce. To address this gap, we hypothesize and test a research model of IS multitasking. Contrary to prior research, we find that IS multitasking yields significant positive outcomes. Our study has implications for attitudes toward IS multitasking, designing training activities, and multitasking systems.