A Neural-Network-Based Approach to Optical Symbol Recognition

In this paper we propose a neural-network-based approach to solving optical symbol recognition problems, from node head recognition to handwritten digit recognition. We demonstrated that node heads could be easily recognized by using a set of fuzzy rules extracted from the parameters of trained neural networks. For handwritten digit recognition we demonstrated that only 12 features are sufficient to achieve a high recognition rate. Several databases were tested to demonstrate the effectiveness and efficiency of the proposed recognition method. This revised version was published online in June 2006 with corrections to the Cover Date.     

神经网络增强学习的梯度算法研究

针对具有连续状态和离散行为空间的Markov决策问题,提出了一种新的采用多层前馈神经网络进行值函数逼近的梯度下降增强学习算法,该算法采用了近似贪心且连续可微的Boltzmann分布行为选择策略,通过极小化具有非平稳行为策略的Bellman残差平方和性能指标,以实现对Markov决策过程最优值函数的逼近,对算法的收敛性和近似最优策略的性能进行了理论分析,通过Mountain-Car学习控制问题的仿真研究进一步验证了算法的学习效率和泛化性能。     

A Neural-Network-Based Approach to Detecting Hyperellipsoidal Shells

This paper presents a novel class of neural networks which can be trained in an unsupervised manner to detect a mixture of hyperellipsoidal shells and/or segments of hyperellipsoidal shells. This approach is computationally and implementationally simpler than other clustering algorithms that have been suggested for this purpose. Experimental results on several data sets are presented.     

Deformable Contour Method: A Constrained Optimization Approach

In this paper, a class of deformable contour methods using a constrained optimization approach of minimizing a contour energy function satisfying an interior homogeneity constraint is proposed. The class is defined by any positive potential function describing the contour interior characterization. An evolutionary strategy is used to derive the algorithm. A similarity threshold T _v can be used to determine the interior size and shape of the contour. Sensitivity and significance of T _v and (a spreadness measure) are also discussed and shown. Experiments on noisy images and the convergence to a minimum energy gap contour are included. The developed method has been applied to a variety of medical images from CT abdominal section, MRI image slices of brain, brain tumor, a pig heart ultrasound image sequence to visual blood cell images. As the results show, the algorithm can be adapted to a broad range of medical images containing objects with vague, complex and/or irregular shape boundary, inhomogeneous and noisy interior, and contour with small gaps.     

一种基于粒子群优化的多目标优化算法

论文提出了一种基于粒子群的多目标优化算法,该算法采用Pareto支配关系来更新粒子的个体最优值和局部最优值,用存储池保存搜索过程中发现的非支配解;采用聚类算法裁剪非支配解,以保持解的分布性能;采用动态惯性权重法来平衡粒子群对解空间的局部搜索和全局搜索,以提高算法的全局收敛性能。实验结果表明,论文算法是有效的,能有效的求解多种多目标优化问题。     

Neural-Network-Based Numerical Control for Milling Machine

We describe a device which uses a neural network to generate part-programs for milling, drilling and similar operations on machining centres, on the basis of 2D, 2.5D or 3D geometric models of prismatic parts, without operator intervention. The neural network consists of networks for prediction of milling strategy, for prediction of surface quality and for the optimisation of technological parameters in milling. We introduce the surface complexity index (SCI) for identifying surfaces which are very difficult to machine. The SCI takes the surface roughness and machining strategy into account. Teaching and testing of the NN is described. The device, which can be retrofitted to a CNC controller, can be trained from a set of typical parts and will then generate new NC part-programs. A case study of a tool used in the automotive supplier industry shows how a milling strategy is proposed, according to set constraints.     

一种用于求解TSP问题的混沌优化算法

针对TSP问题,提出了一种用于解决组合优化问题的混沌优化算法。在寻优过程中,对解矩阵不断进行“交换”和“移位”两种操作,利用混沌搜索方法确定所进行操作的矩阵位置。仿真结果表明,该方法的搜索效率要明显高于其他的优化方法。同时,寻优结果对参数的变化不很敏感。     

一种解决非线性优化问题的新方法

本文对组合优化领域中出现的新型随机搜索方法——蚂蚁算法进行了引伸和扩展，得出一种可以快速有效地解决无约束和有约束非线性优化问题的新途径。利用一系列的典型函数进行测试，证实其全局收敛性好，运算速度较快，具有良好的实用性能。     

Neural-Network-Based Contact Force Observers for Haptic Applications

In the majority of robotic and haptic applications, including manipulation and human-robot interaction, contact force needs to be monitored and controlled. Transparent implementation of bilateral teleoperation or haptic controllers necessitates the exchange of operator and environment contact forces. This requires the use of expensive commercially available force/torque sensors, which are rather bulky, are vulnerable to impact forces, and increase system inertia and compliance. An alternative solution is the use of dynamic force observers, which estimate external forces using system dynamic model. However, due to the uncertainties in system dynamic structure and parameters, these model-based observers do not produce accurate force estimates, and often create a dynamic lag that may cause bandwidth limitation and instability. This paper proposes two neural-network-based force/torque observers that do not require a system dynamic model. The observers can estimate human hand force and environment contact force with up to 98.3% accuracy in the sense of mean-square error, and with negligible dynamic lag. The performance of the proposed observers are extensively analyzed in separate human-robot and robot-environment experimental settings, and in a two-channel bilateral teleoperation control loop with multiple runs with two Planar Twin-Pantograph haptic devices     

EOFDM:一种面向众核架构的最低能耗搜索方法

面向能耗优化的面积(核数)-功率(频率)分配问题是当前众核处理器研究热点之一.通过性能-功耗模型了解其在核数-频率空间的分布规律,然后在核数和频率级别这2个维度上通过实测执行逐步搜索,可以获取“核数-频率”配置的最优解,从而达到能耗优化的目的;然而本领域现有方法在核数-频率空间内实测搜索最低能耗时收敛速度慢、搜索开销大、可扩展性差.针对此问题,提出了一种基于求解最优化问题的经典数学方法——可行方向法的最低能耗搜索方法(energy-efficient optimization based on feasible direction method,EOFDM),每次执行都能从核数和频率2个维度上同时减小搜索空间,在迭代执行中快速收敛至最低能耗点.该方法与现有研究中最优的启发式爬山法(hill-climbing heuristic,HCH)进行了对比实验,平均执行次数、执行时间和能耗分别降低39.5％,46.8％,48.3％,提高了收敛速度,降低了搜索开销;当核数增加一倍时,平均执行次数、执行时间和能耗分别降低48.8％,51.6％,50.9％;当频率级数增加一倍时,平均执行次数、执行时间和能耗分别降低45.5％,49.8％,54.4％,在收敛速度、搜索开销和可扩展性方面均有提高.     

一种新的优化方法及其应用

本文分析了近年来出现的一些优化算法，如模拟退火法、遗传算法和趋化性算法等的特点。在此基础上，针对连续优化命题，提出了一种新的算法，并以一神经网络参数学习为例说明了该算法的有效性     

基于全局优化的影象变形方法

该文首先对影象计算中变形算法的研究作了一个简要的概述，然后结合我们目前所进行的智能动画的研究工作，提出了一个在影象空间中基于全局优化的影象变形算法。在文中第二节较主详细地介绍了该算法中所用到的散乱点播值算法，最后给出了该算法的详细描述。     

支持向量机参数优化的一种新方法

支持向量机(SVM)的性能与SVM参数的选择有关.SVM参数的优化需要一个准则,本文提出了一种以原空间中样本到分类面的最短代数距离最大为准则的SVM参数优化方法.该方法旨在使SVM分类面在原空间中使样本"平分秋色",更能体现SVM分类器的结构风险最小化的原则.算法简单、几何直观性好、易于实现.通过在双螺旋线样本和Iris样本集上所作测试证明了该方法的有效性.     

一种新的遗传混沌优化组合方法

在分析了遗传算法与混沌优化方法的优缺点的基础上, 提出了一种新的遗传混沌优化组合方法. 该算法能克服混沌优化在大范围内失效的缺点, 并能提高遗传算法的局部搜索能力和搜索精度. 同时证明该算法能以概率 1收敛到全局最优值. 应用该方法对 6个测试函数进行优化计算得到了比较满意的结果.     

一种新颖的软件可控Cache优化方法

由于Cache污染问题,传统的仅由硬件控制的Cache替换策略不能得到令人满意的Cache利用率。随着软件可控Cache机制的出现,编译器开始可以直接控制Cache替换,改善Cache行为。本文证明了一个Cache提示优化定理,并依该定理提出了一个由编译器辅助控制的Cache替换策略:最优Cache划分(OCP)。OCPCache替换策略简化了Cache行为和Cache失效分析方法。实验结果表明OCPCache替换策略能有效地降低Cache失效率。     

间歇生产过程优化调度的新方法

本文提出了间歇生产过程在线生产调度算法的评价指标，该指标较好地综合了间歇过程等待时间和操作单元闲置时间对经济效益的影响；并提出基于预测的短期生产调度在线校正算法ＰＯＭＡ，改进了以往算法的“近视“效应。仿真结果表明，ＰＯＭＡ算法具有很好的效果和很强的鲁棒性。该算法原理简单，对间歇工业生产过程优化调度具有较大的实用价值。     

一种单交叉路口的信号混合优化方法

针对包含不等式约束和等式约束的城市单交叉路口信号优化问题,为缓解交通堵塞和安全性,设计了一种混合优化方法.方法首先采用自适应惩罚策略,将具有不等式约束和等式约束的优化问题转变为仅包含决策变量上、下限约束的优化问题;然后再分别采用自适应实数编码遗传算法和一种变搜索空间局部搜索算法进行混合优化,为了提高实数编码遗传算法的优化效果,设计了一种自适应交叉概率和变异概率.最后针对多种交通需求模式,应用混合优化方法进行了大量的仿真计算,结果表明在城市单交叉路口信号优化问题中具有良好的优化效果.     

A Pipelining Loop Optimization Method for Dataflow Architecture

与 exascale 来超级计算的时代,电源效率成为了最重要的障碍造一个 exascale 系统。Dataflow 建筑学在为科学应用完成高电源效率有本国的优点。然而,最先进的 dataflow 体系结构没能为循环处理利用高并行。处理这个问题,我们建议一个 pipelining 环优化方法(PLO ) ,它在处理元素(PE ) 在环流动做重复 dataflow 的数组加速器。这个方法由二种技术,帮助建筑学的硬件重复和帮助说明的软件重复组成。在硬件重复执行模型,一个在薄片上循环控制器被设计产生循环索引,减少计算内核并且打为 pipelining 执行的一个好基础的复杂性。在软件重复实行模型,另外的环指令被论述解决重复相关性问题。经由这二种技术,准备好了每周期执行的指令的平均数字被增加使浮点联合起来忙。当这二种技术的硬件费用是可接受的时,模拟结果证明分别地,我们的建议方法平均由 2.45x 和 1.1x 在浮点效率超过静电干扰和动态循环执行模型。     

基于协同进化博弈的多学科设计优化

复杂系统的设计问题可以非层次分解为并行的多个子空间优化设计问题。多学科优化的迭代过程可看成子空间博弈的过程。各冲突子目标协商一致条件下,子空间合作博弈的均衡点能达成原系统的整体最优,并给出协同进化算法求解博弈的Nash均衡点的计算框架。以某型民用客机的总体优化设计为例,将其分解成气动和重量两个子空间优化。设计变量不重叠地分布于各子空间,两冲突子目标分配相同权值,线性加权组合而形成的单目标作为各子空间共同的优化目标。计算结果表明此方法是有效的。     

A Modified Quasi-Newton Method for Optimization in Simulation

Optimization in Simulation is an important problem often encountered in system behavior investigation; however, the existing methods such as response surface methodology and stochastic approximation method are inefficient. This paper presents a modification of a quasi-Newton method, in which the parameters are determined from some numerical experiments. To demonstrate the validity of the devised method, two examples resembling the M/M/1 queueing problem are solved. The closeness of the converged solutions to the optimal solutions and a comparison with two stochastic approximation methods indicate that the modified quasi-Newton method as devised in this paper is a robust and efficient method for solving optimization problems in simulation.