统计启发式搜索算法在函数优化中的应用

本文讨论统计启发式搜索算法在优化计算中的应用，主要是函数求优化。为此引入新的ＭＡＸ统计量，构造相应的ＳＡ算法（称之为ＳＡ（ＭＡＸ）算法），并分析了新算法的精度和计算复杂性。最后给出计算机模拟的结果，以验证理论的正确性。     

新的图像边界检测的快速递推算法

本文推荐一种新的图像边界检测的快速算法，通过ＣＡＮＮＹ的边界检测函数理论，我们证明了这种新的边界检测函数在信噪比（ＳＩＧＮＡＬＴＯＮＯＩＳＥＲＡＴＩＯ）边界检测精度（ＬＯＣＡＬＩＺＡＴＩＯＮ）和伪边界平均距离（ＭＵＬＴＩＰＬＥＲＥＳＰＯＮＳＥ）这三个性能指标上都优于目前已知的任何一种边界检测函数，本文首先给出一维信号的递推公式，然后再应用到三维图像处理中，该检测函数的特点是计算简单，对于一维信号的     

一类GASA混合策略及其收敛性研究

结合模拟退火算法（ＳＡ）和遗传算法（ＧＡ）提出一类ＧＡＳＡ混合优化策略，借助于非平稳马氏链理论证明混合算法的全局渐近收敛性，同时实性地分析了算法的优化效率。     

带有变量中误差的Logit回归模型及其计算

带有变量中误差的Ｌｏｇｉｔ回归模型及其计算吕纯濂，陈舜华（南京气象学院）Ｈ．Ｋｕｃｈｅｎｈｏｆｆ（慕尼黑大学统计与科学理论研究所）ＴＨＥＬＯＧＩＴＲＥＧＲＥＳＳＩＯＮＭＯＤＥＬＷＩＴＨＥＲＲＯＲＳ－ＩＮ－ＶＡＲＩＡＢＬＥＳＡＮＤＩＴＳＣＡＬＣＵＬＡＴ...     

一种LDA与SVM混合的多类分类方法

针对决策有向无环图支持向量机（ＤＤＡＧＳＶＭ）需训练大量支持向量机（ＳＶＭ）和误差积累的问题,提出一种线性判别分析（ＬＤＡ）与ＳＶＭ 混合的多类分类算法．首先根据高维样本在低维空间中投影的特点,给出一种优化ＬＤＡ 分类阈值;然后以优化ＬＤＡ 对每个二类问题的分类误差作为类间线性可分度,对线性可分度较低的问题采用非线性ＳＶＭ 加以解决,并以分类误差作为对应二类问题的可分度;最后将可分度作为混合ＤＤＡＧ 分类器的决策依据．实验表明,与ＤＤＡＧＳＶＭ 相比,所提出算法在确保泛化精度的条件下具有更高的训练和分类速度．     

模拟退火算法与遗传算法的结合

模拟退火算法与遗传算法的结合王雪梅，王义和（哈尔滨工业大学计算机科学与工程系哈尔滨１５０００１）ＴＨＥＣＯＭＢＩＮＡＴＩＯＮＯＦＳＩＭＵＬＡＴＥＤＡＮＮＥＡＬＩＮＧＡＮＤＧＥＮＥＴＩＣＡＬＧＯＲＩＴＨＭＳ￥ＷＡＮＧＸｕｅｍｅｉ；ＷＡＮＧＹｉｈｅ（Ｄｅ...     

基于类比的学习式搜索算法AMO.GLSA

本文首先给出了学习式搜索的一个问题模型，然后（５）中ＧＬＳ搜索解题系统的基础上，本文描述了一个多目标学习搜索算法ＭＯ．ＧＬＳＡ，并对该算法作出一性能评价，最后，文中给出了一个基于类比的学习搜索算法ＡＭＯ．ＧＬＳＡ。     

块角型约束线性规划问题的内点分解算法

块角型约束线性规划问题的内点分解算法吴力（中国科学院计算数学与科学工程计算研究所）ＡＤＥＣＯＭＰＯＳＩＴＩＯＮＡＬＧＯＲＩＴＨＭＦＯＲＬＩＮＥＡＲＰＲＯＧＲＡＭＭＩＮＧＰＲＯＢＬＥＭＳＷＩＴＨＢＬＯＣＫＡＮＧＵＬＡＲＣＯＮＳＴＲＡＩＮＴＳ￥ＷｕＬｉ（...     

经验点滴

显示卡知识大全 （接上期） ＳＧＲＡＭ（Ｓｙｎｃｈｒｏｎｏｕｓ Ｇｒａｐｈｉｃｓ ＲＡＭ） ＳＧＲＡＭ（同步）是一种比较新的显存，而且它是为专为显示卡所设计的，它改进了过去低效能显存传输率较低的缺点，为显示卡性能的提高创造了条件。但是因为其设计制造成本过高，在普通显卡上采用的较少，一般都是运用在高端加速卡上。现在有很多低档３Ｄ加速卡都使用ＳＧＲＡＭ，但是经过比较你会发现其性能甚至还不如使用ＳＤＲＡＭ的同等产品。 ＳＤＲＡＭ（Ｓｙｎｃｈｒｏｎｏｕｓ ＤＲＡＭ）相信大家对这种显存并不陌生，ＳＤＲＡＭ与早期…     

一种LDA与SVM混合的多类分类方法

针对决策有向无环图支持向量机（ＤＤＡＧＳＶＭ）需训练大量支持向量机（ＳＶＭ）和误差积累的问题,提出一种线性判别分析（ＬＤＡ）与ＳＶＭ 混合的多类分类算法．首先根据高维样本在低维空间中投影的特点,给出一种优化ＬＤＡ 分类阈值;然后以优化ＬＤＡ 对每个二类问题的分类误差作为类间线性可分度,对线性可分度较低的问题采用非线性ＳＶＭ 加以解决,并以分类误差作为对应二类问题的可分度;最后将可分度作为混合ＤＤＡＧ 分类器的决策依据．实验表明,与ＤＤＡＧＳＶＭ 相比,所提出算法在确保泛化精度的条件下具有更高的训练和分类速度．

     

Application of heuristic and hybrid-GASA algorithms to tool-path optimization problem for minimizing airtime during machining

In this paper, heuristic algorithms such as simulated annealing (SA), genetic algorithm (GA) and hybrid algorithm (hybrid-GASA) were applied to tool-path optimization problem for minimizing airtime during machining. Many forms of SA rely on random starting points that often give poor solutions. The problem of how to efficiently provide good initial estimates of solution sets automatically is still an ongoing research topic. This paper proposes a hybrid approach in which GA provides a good initial solution for SA runs. These three algorithms were tested on three-axis-cartesian robot during milling of wood materials. Their performances were compared based on minimum path and consequently minimum airtime. In order to make a comparison between these algorithms, two cases among the several milling operations were given here. According to results obtained from these examples, hybrid algorithm gives better results than other heuristic algorithms alone. Due to combined global search feature of GA and local search feature of SA, hybrid approach using GA and SA produces about 1.5% better minimum path solutions than standard GA and 47% better minimum path solutions than standard SA.     

Glass container production scheduling through hybrid multi-population based evolutionary algorithm

Driven by a real-world application in the capital-intensive glass container industry, this paper provides the design of a new hybrid evolutionary algorithm to tackle the short-term production planning and scheduling problem. The challenge consists of sizing and scheduling the lots in the most cost-effective manner on a set of parallel molding machines that are fed by a furnace that melts the glass. The solution procedure combines a multi-population hierarchically structured genetic algorithm (GA) with a simulated annealing (SA), and a tailor-made heuristic named cavity heuristic (CH). The SA is applied to intensify the search for solutions in the neighborhood of the best individuals found by the GA, while the CH determines quickly values for a relevant decision variable of the problem: the processing speed of each machine. The results indicate the superior performance of the proposed approach against a state-of-the-art commercial solver, and compared to a non-hybridized multi-population GA.     

A new optimization method: Big Bang–Big Crunch

Nature is the principal source for proposing new optimization methods such as genetic algorithms (GA) and simulated annealing (SA) methods. All traditional evolutionary algorithms are heuristic population-based search procedures that incorporate random variation and selection. The main contribution of this study is that it proposes a novel optimization method that relies on one of the theories of the evolution of the universe; namely, the Big Bang and Big Crunch Theory. In the Big Bang phase, energy dissipation produces disorder and randomness is the main feature of this phase; whereas, in the Big Crunch phase, randomly distributed particles are drawn into an order. Inspired by this theory, an optimization algorithm is constructed, which will be called the Big Bang–Big Crunch (BB–BC) method that generates random points in the Big Bang phase and shrinks those points to a single representative point via a center of mass or minimal cost approach in the Big Crunch phase. It is shown that the performance of the new (BB–BC) method demonstrates superiority over an improved and enhanced genetic search algorithm also developed by the authors of this study, and outperforms the classical genetic algorithm (GA) for many benchmark test functions.     

Heuristic methods for vehicle routing problem with time windows

This paper documents our investigation into various heuristic methods to solve the vehicle routing problem with time windows (VRPTW) to near optimal solutions. The objective of the VRPTW is to serve a number of customers within predefined time windows at minimum cost (in terms of distance travelled), without violating the capacity and total trip time constraints for each vehicle. Combinatorial optimisation problems of this kind are non-polynomial-hard (NP-hard) and are best solved by heuristics. The heuristics we are exploring here are mainly third-generation artificial intelligent (AI) algorithms, namely simulated annealing (SA), Tabu search (TS) and genetic algorithm (GA). Based on the original SA theory proposed by Kirkpatrick and the work by Thangiah, we update the cooling scheme and develop a fast and efficient SA heuristic. One of the variants of Glover's TS, strict Tabu, is evaluated and first used for VRPTW, with the help of both recency and frequency measures. Our GA implementation, unlike Thangiah's genetic sectoring heuristic, uses intuitive integer string representation and incorporates several new crossover operations and other advanced techniques such as hybrid hill-climbing and adaptive mutation scheme. We applied each of the heuristics developed to Solomon's 56 VRPTW 100-customer instances, and yielded 18 solutions better than or equivalent to the best solution ever published for these problems. This paper is also among the first to document the implementation of all the three advanced AI methods for VRPTW, together with their comprehensive results.     

基于SA的改进遗传算法的测试数据生成研究

测试数据的自动生成研究是软件测试的一个焦点问题,测试数据的自动生成可以提高测试工作效率,节约测试成本.考虑遗传算法(GA)和模拟退火算法(SA)各自优缺点,提出遗传/模拟退火(GASA)混合算法的策略,在标准的GA中融入SA,在GA的局部搜索中引入SA,SA的随机状态受限于遗传优化算法的结果,GA的种群更新是由SA的退温算法和随机状态产生函数来控制,从而得到最优解.GA-SA算法取长补短,提高了算法的全局和局部搜索能力,能避免GA过早收敛,提高了算法搜索最优解的能力.实验结果表明,GASA算法寻找最优解所需的迭代次数明显优于标准GA.     

A New Heuristic Search Technique-Algorithm SA

In this paper, we present a new heuristic searching algorithm by introducing the statistical inference method on the basis of algorithm A (or A*). It is called algorithm SA. In a simplified search space, a uniform m-ary tree, we obtain the following result. Using algorithm SA, a goal node can be found with probability one, and its mean complexity is O(N·ln N) where N is the depth at which the goal is located.     

The hybrid heuristic genetic algorithm for job shop scheduling

Scheduling for the job shop is very important in both fields of production management and combinatorial optimization. However, it is quite difficult to achieve an optimal solution to this problem with traditional optimization methods owing to the high computational complexity (NP-hard). Genetic algorithms (GA) have been proved to be effective for a variety of situations, including scheduling and sequencing. Unfortunately, its efficiency is not satisfactory. In order to make GA more efficient and practical, the knowledge relevant to the problem to be solved is helpful. In this paper, a kind of hybrid heuristic GA is proposed for problem n/m/G/C_max, where the scheduling rules, such as shortest processing time (SPT) and MWKR, are integrated into the process of genetic evolution. In addition, the neighborhood search technique (NST) is adopted as an auxiliary procedure to improve the solution performance. The new algorithm is proved to be effective and efficient by comparing it with some popular methods, i.e. the heuristic of neighborhood search, simulated annealing (SA), and traditional GA.     

基于遗传模拟退火算法的门阵列布局方法

为实现门阵列模式布局,将遗传算法与模拟退火算法相结合,提出一种新的遗传模拟退火算法,利用遗传算法进行全局搜索,利用模拟退火法进行局部搜索,在进化过程中采用精英保留策略,对进化结果进行有选择的模拟退火操作,既加强了局部搜索能力又防止陷入局部最优。实验结果表明,与传统遗传算法相比,该算法能够有效提高全局搜索能力。     

Data Generation for Path Testing

We present two stochastic search algorithms for generating test cases that execute specified paths in a program. The two algorithms are: a simulated annealing algorithm (SA), and a genetic algorithm (GA). These algorithms are based on an optimization formulation of the path testing problem which include both integer- and real-value test cases. We empirically compare the SA and GA algorithms with each other and with a hill-climbing algorithm, Korel's algorithm (KA), for integer-value-input subject programs and compare SA and GA with each other on real-value subject programs. Our empirical work uses several subject programs with a number of paths. The results show that: (a) SA and GA are superior to KA in the number of executed paths, (b) SA tends to perform slightly better than GA in terms of the number of executed paths, and (c) GA is faster than SA; however, KA, when it succeeds in finding the solution, is the fastest.