演化多目标优化中的几何热力学选择

热力学遗传算法(Thermodynamical Genetic Algorithms,TDGAs)借鉴热力学中的自由能极小过程来统一处理多目标优化在逼近性和多样性两方面的任务.为提高TDGA的运行效率和解集分布均匀性,提出了一种几何热力学选择.在该选择中首先定义角度熵通过扇形采样来度量种群逼近方向的多样性.然后利用距离精英定义距离能量来度量种群的逼近程度,避免了耗时的非劣分层操作.此外,引入分量热力学替换规则以较低计算代价驱动种群的几何自由能快速下降.在多目标0/1背包问题上的实验结果表明,几何热力学选择极大地提高了TDGA的运行效率和解集分布均匀性;采用该选择的TDGA算法可生成与NSGA-II在逼近性和分布多样性上性能相当的解,但在运行效率上明显优于NSGA-II.     

基于最小熵产生选择策略的遗传算法研究

针对遗传算法(Genetic Algorithm, GA)容易陷入局部最优的问题,借鉴热力学非平衡定态下的最小熵增原理,提出一种基于最小熵增原理的热力学选择策略,使个体的选择不再完全依赖于适应值。通过最小熵产生选择策略使种群在保证收敛速度的同时保持多样性,有效避免了种群陷入局部最优。通过定义个体密度来度量种群多样性,利用精英策略驱动种群熵产生快速下降;当种群多样性过低时,使用基于最小熵产生的选择策略产生新种群以保证种群多样性。在0/1背包问题和数值测试问题上的实验结果均表明,该策略能很好地保证解集分布的均匀性,防止种群陷入局部最优。同时,该策略也可应用于目前较新改进的遗传算法中,对算法效率也有一定的改进,具有很好地普适性。     

基于GrC的信息系统软规则及其度量关系的研究

粒计算(GranularComputing,简称GrC)是一种新的软计算方法。该文利用信息颗粒的位表示(BitRepresenta-tions)来进行信息系统软规则及其度量之间关系的研究。具体地,首先利用软规则对关联规则、决策规则、函数依赖之间的关系进行了分析,然后对关联规则度量、决策规则度量、外延的函数依赖度量的关系进行了研究,并且建立了这些度量的统一模型。     

基于决策熵的决策树规则提取方法

在决策表中,决策规则的可信度和对象覆盖度是衡量决策能力的重要指标。以知识粗糙熵为基础,提出决策熵的概念,并定义其属性重要性;然后以条件属性子集的决策熵来度量其对决策分类的重要性,自顶向下递归构造决策树;最后遍历决策树,简化所获得的决策规则。该方法的优点在于构造决策树及提取规则前不进行属性约简,计算直观,时间复杂度较低。实例分析的结果表明,该方法能获得更为简化有效的决策规则。     

基于决策熵的决策树规则提取方法

在决策表中,决策规则的可信度和对象覆盖度是衡量决策能力的重要指标。以知识粗糙熵为基础,提出决策熵的概念,并定义其属性重要性;然后以条件属性子集的决策熵来度量其对决策分类的重要性,自顶向下递归构造决策树;最后遍历决策树,简化所获得的决策规则。该方法的优点在于构造决策树及提取规则前不进行属性约简,计算直观,时间复杂度较低。实例分析的结果表明,该方法能获得更为简化有效的决策规则。     

用于约束多目标优化问题的热力学遗传算法

为了保持所求得的约束多目标优化问题Pareto最优解的适应度与多样性,在NSGA-Ⅱ基础上提出了一种用于求解有约束的多目标优化问题的热力学遗传算法.结合热力学中自由能与熵的概念,利用热力学中熵与能量的竞争来保持种群的适应度与多样性的平衡,设计了热力学算子.根据非支配排序Pareto分层结构建立分层小生境来改进选择算子,弥补了选择算子不足.实验结果表明:该算法不仅得到的解在空间分布均匀,收敛性好,同时解集具有较广的分布空间.     

基于混合差异度控制的基因表达式编程

基因表达式编程( GEP)是一种进化算法,存在局部极小问题,解决此问题的一般方法是保持进化过程中种群的差异度.为了保证进化过程中种群的差异度,文中提出一种融合种群空间和样本空间的种群差异度度量方法.并基于此融合种群差异度度量方法,提出差异控制的GEP进化算法.同时在初始种群生成时,针对GEP结构的特殊性,将敌手理论应用于GEP种群初始化.实验结果表明文中算法能较有效避免过早陷入局部极小.     

决策信息系统不确定性的熵度量

提出基于Rough集信息论观点的决策表不确定性的熵度量指标,相比代数论观点的决策表不确定性度量指标在含义上更清晰,对不确定性的度量更全面.运用不同的熵度量指标,可分别度量一致性决策信息系统和不一致性决策信息系统的规则前件不确定性、规则后件不确定性、规则整体随机性以及决策表整体不确定性,并对同一决策信息系统运用不同的熵度量指标对其不确定性进行分析.     

基于共享机制的自适应混合遗传算法

共享机制小生境遗传算法常由于保持算法种群的多样性而减缓了全局收敛速度.针对共享机制的这个缺陷,提出了一种基于共享机制的自适应混合遗传算法.将熵的概念引入共享机制,提出了用以度量种群多样性的小生境熵的概念;构造了小生境半径和进化参数(交叉、变异概率)的自适应计算方法;设计了用于增强算法局部搜索寻优能力的扩展突变算子.最后实验表明,该算法对于解决多模态函数优化问题具有很好的全局搜索能力和较快的收敛速度,能够有效避免早熟收敛.     

基于决策熵的不完备信息系统的知识约简方法

为有效地解决不完备信息系统的知识约简,得到更优的决策规则集,研究了基于容差关系的决策熵在不完备信息系统中能客观反映决策规则的决策能力,提出了一种基于决策熵的不完备知识约简方法.该方法基于决策熵的思想,考虑了决策规则可信度和对象覆盖度,同时引入了容差关系,以决策熵的属性重要性度量为启发信息进行知识约简,最终得到确定的规则集.仿真实验结果表明了该算法的可行性.     

二维直方图准分的Renyi熵快速图像阈值分割

针对传统二维Renyi熵(RE)分割法分割结果不够准确和计算复杂度高的问题,提出一种快速的二维RE准分法.首先,用与主对角线平行的四条斜线将直方图分成内点区、边界点区和噪声点区,并对噪声点区进行去噪处理以便获得更好的分割性能.然后,对内点区与边界点区在RE公式中的对应量准确取值使阈值选取更准确.最后,提出二维RE准分法的一般递推算法,并在此算法的基础上利用RE在二维直方图上的计算特性和两个公式导出快速的二维RE阈值选取算法来降低计算复杂度.实验结果表明,与对比方法相比,文中方法不仅分割更准确和抗噪性更强,而且其运行时间少,与二维RE斜分法运行时间相近.     

Large-scale structural optimization using metaheuristic algorithms with elitism and a filter strategy

Large-scale structural optimization often requires numerous finite element analyses to assess the feasibility of the derived solutions during the optimization process, which consume most of the computational cost. To enhance the computational efficiency, this study introduces a filter strategy aiming to eliminate the redundant constraint violation evaluations in large-scale structural optimization using metaheuristic algorithms. Based on the solution selection rule, this study separates the metaheuristic algorithms into two categories: replacement and elitism. The filter mechanism founds on elitism of the metaheuristic algorithms and reduces substantially the number of structural analyses without compromising the effectiveness of the optimization algorithms and the constraint handling techniques. This study also defines a parameter, R, to assess the enhancement performance of the computational efficiency improved by the proposed method. Results from both mathematical simulations and two large-scale structural optimization examples using various metaheuristic algorithms demonstrate that the harmony search (HS) leads always to the lowest R value. The R value is less than 0.4 and is even as small as 0.09 for the 942-bar example, which means over 90% of time savings compared with the penalty method and the Deb rule and the quality of the final optimum also does not depend on the value of R.     

Comparative study of four penalty-free constraint-handling techniques in structural optimization using harmony search

This study investigates the search capability, stability, and computational efficiency of four improved penalty-free constraint-handling techniques (CHTs), including the death penalty, the Deb rule, the filter method, and the mapping strategy, in structural optimization using harmony search (HS). The first three general-purpose CHTs have been improved by hybridizing with a structural analysis filter strategy to enhance their computational efficiency based on the characteristics of structural optimization and the solution updating rule of the HS. This study also has modified the mapping operator of the mapping strategy to handle size and shape optimization with Euler buckling constraints. Four numerical examples examine the performances of these CHTs. The comparative results show that the mapping strategy exhibits apparent superiority both in search capability and stability among the four. However, it also demands the most computational cost, and the improved Deb rule becomes the most competitive method while considering computational efficiency. The difference between the death penalty and the Deb rule method is that the feasible solution initialization process required by the death penalty method will deteriorate its computational efficiency in problems with small feasible space. The filter method always reserves some infeasible solutions to guide the search in the iteration process. However, its performances are inferior to the other three approaches in four benchmark problems.

     

基于代理模型的差分进化约束优化

针对耗时计算目标函数的约束优化问题,提出用代理模型来代替耗时计算目标函数的方法,并结合目标函数的信息对约束个体进行选择,从而提出基于代理模型的差分进化约束优化算法。首先,采用拉丁超立方采样方法建立初始种群,用耗时计算目标函数对初始种群进行评估,并以此为样本数据建立目标函数的神经网络代理模型。然后,用差分进化方法为种群中的每一个亲本产生后代,并对后代使用代理模型进行评估,采用可行性规则来比较后代与其亲本并更新种群,根据替换机制将种群中较劣的个体替换为备用存档中较优的个体。最后,当达到最大适应度评估次数时算法停止,给出最优解。该算法与对比算法在10个测试函数上运行的结果表明,该算法得出的结果更精确。将该算法应用于工字梁优化问题的结果表明,相较于优化前的算法,该算法的适应度评估次数减少了80%;相对于FROFI（Feasibility Rule with the incorporation of Objective Function Information）算法,该算法的适应度评估次数减少了36%。运用所提算法进行优化可以有效减少调用耗时计算目标函数的次数,提升优化效率,节约计算成本。     

A novel decomposition-based ensemble model for short-term load forecasting using hybrid artificial neural networks

Highly accurate short-term load forecasting (STLF) is essential in the operation of power systems. However, the existing predictive methods cannot achieve an effective balance between prediction accuracy and computational cost. Furthermore, the prediction residual is rarely used to improve the predictive accuracy in STLF. This paper proposes a novel decomposition-based ensemble model for the STLF task. First, an optimized empirical wavelet transform (OEWT) is developed to rationally decompose the STLF load by combining the approximate entropy method with the empirical wavelet transform. Particularly, OEWT improves both prediction accuracy and computational cost in STLF. Second, a new hybrid machine learning method (named master learner) is proposed by rationally combining long short-term memory networks (LSTMs) with broad learning system (BLS) in STLF, effectively strengthening the predictive accuracy without significantly increasing the computational cost. Third, a residual learning model (named residual learner) is developed in the master learner to extract the effective predictive information from residual results, further improving the prediction accuracy in STLF. Fourth, an auxiliary learner is proposed by introducing another BLS to connect the input and output of the proposed model, enhancing the predictive robustness. The proposed decomposition-based ensemble model is compared with state-of-the-art and traditional models in STLF. Experimental results show that the model not only has high predictive accuracy and robustness but also low computational cost.

     

基于信息熵的集成学习过程多样性度量研究

基分类器的多样性是提升集成学习的精度和泛化能力的重要因素,大数据环境下的传统后验证多样性度量方法计算效率较低,提出一种基于信息熵的过程多样性度量方法。通过使用分类器各属性的增益及其所在树层次得到属性集的联合增益,并计算分类器间的熵距离评估其多样性,利用熵距离按照K-means方法即可动态购置集成学习分类器。在西瓜数据集和典型分类数据集上进行比较研究,发现与传统集成学习方法相比,该方法具有相近的准确性和更高的计算效率。     

改进的混合蛙跳算法及其在多阈值图像分割中的应用

针对混合蛙跳算法(Shuffled Frog Leaping Algorithm,SFLA)存在的计算复杂度高、优化效率不理想等问题,提出了一种改进的混合蛙跳算法(Improved Shuffled Frog Leaping Algorithm,ISFLA)。在原始 SFLA的基础上进行如下改进:首先,将其中每次只更新组内最差青蛙的方式改为更新组内所有青蛙的方式,这既增大了获得优质解的概率,又省去了调整组内迭代次数的步骤,从而提升了优化效率和可操作性;其次,将基于局部最优更新的方法和基于全局最优更新的方法融合为一种混合扰动更新方法,从而避免了复杂条件的选择步骤,进一步提升了优化效率;最后,去掉随机更新方式,以免优质解被破坏,从而提高了整体的优化性能。将 ISFLA 用于 CEC2005和CEC2015连续基准函数的优化测试和基于Renyi 熵的灰度和彩色图像分割的多阈值选择实验中,结果表明,与 SFLA 和state-of-the-art的LSFLA 相比,ISFLA 具有更高的优化效率,更适用于多阈值图像分割的阈值选择。     

面向视频冷启动问题的点击率预估

视频的点击率预估是视频推荐系统中的重要任务之一,推荐系统可以根据点击率的预估调整视频推荐顺序以提升视频推荐的效果.近年来,随着视频数量的爆炸式增长,视频推荐的冷启动问题也变得愈发严重.针对这个问题,提出了一个新的视频点击率预估模型,通过使用视频的内容特征以及上下文特征来加强视频点击率预估的效果;同时,通过对冷启动场景的模拟训练和基于近邻的替代方法提升模型应对新视频点击率预估的能力.提出的模型可以同时对旧视频和新视频进行点击率预估.在两个真实的电视剧（Track_1_series）和电影（Track_2_movies）点击率预估数据集上的实验表明：提出的模型可以显著改善对旧视频的点击率预估性能,并在两个数据集上均超过了现有的模型;对于新视频,相比于不考虑冷启动问题的模型只能获得0.57左右的AUC性能,该模型在两个数据集上分别获得0.645和0.615的性能,表现出针对冷启动问题更好的鲁棒性.     

Least-squares parameter estimation of linear systems with noisy input–output data

This paper studies the computational efficiency of the bias-eliminated least-squares (BELS) method recently proposed for estimating linear systems in the presence of input and output noises. It is found that the BELS method based on expanding the denominator polynomial of the system transfer function by two dimensions may involve some redundant computations due to its handling of an augmented system model in its estimation scheme. To improve the computational efficiency, a direct estimation scheme is proposed to identify the underlying noisy input–output system. Numerical results show that the computational cost can be considerably reduced using such a new estimation scheme.