上下文一致性错误及其处理对应用的影响

在普适计算环境中,上下文信息由于受到噪声等不可控因素的干扰,面临着一致性错误的问题。这些错误会影响应用的正常运行,使其表现异常甚至失效。目前已经有自动的一致性错误处理方法来保障应用所获取的上下文满足数据层面的一致性约束,然而在数据层面保持一致是否能提升应用的运行质量并不明确。系统地研究了这个问题,并基于一组真实的上下文感知机器人小车应用,设计了一系列受控实验,来分析和比较在模拟和真实世界中上下文一致性错误的处理方式对应用运行质量的改善程度及潜在的负面影响。利用该实验的分析结果,将有助于提高普适计算中上下文感知应用的运行质量。     

Hybrid-Fixing:上下文一致性错误的正确修复

在普适计算中,上下文持续快速变化,上下文感知应用根据上下文变化自动调整自身的行为以作出适应.然而,由于不可预测和控制的环境噪声以及环境动态变化等诸多因素的影响,环境上下文会发生一致性错误,从而导致应用表现异常甚至失效.为了解决这些问题,上下文一致性错误需要被自动并正确地修复,现基于已有工作提出了一项新的修复技术hybrid-fixing,它结合了对一致性约束的静态分析和修复动作的动态产生,即使一致性约束内部存在复杂依赖关系,也能确保所生成的修复用例必然正确.实验结果表明,这项修复技术大幅提高了一致性约束内部存在复杂依赖关系下一致性错误修复的成功率,并只花费了很小的时间开销.     

一种基于新约束处理方法的遗传算法

针对目前的约束处理方法中存在的问题,提出一种新的约束处理方法。该方法通过可行解和不可行解混合交叉的方法对问题的解空间进行搜索,对可行种群和不可行种群分别进行选择操作。避免了惩罚策略中选取惩罚因子的困难,使得约束处理问题简单化。实例测试结果表明,该约束处理方法的有效性。     

求解约束优化问题的动态邻域粒子群算法*

粒子群算法(PSO)求解约束优化问题存在较严重的早熟收敛现象,为了有效抑制早熟收敛,提出了基于改进的约束自适应方法的动态邻域粒子群算法(IPSO)。算法采用动态邻域策略提高算法的全局搜索能力,设计了一种改进的自适应约束处理方法,根据迭代代数线性增加搜索偏向系数,在早期偏向于搜索可行解,在后期偏向于搜索最优解,并引入序列二次规划增强算法的局部搜索能力。通过基准测试函数实验对比分析,表明该算法对于约束优化问题具有较好的全局收敛性。     

弧一致性符号ADD算法及在CSP求解中的应用

约束满足问题(CSP)是人工智能领域中一个重要的研究课题,弧一致性(AC)技术是提高约束满足问题求解效率的一种有效技术。对传统弧一致性技术进行了改进,给出了弧一致性的符号代数决策图(ADD)算法并将其应用于CSP求解。传统弧一致性技术在压缩问题的搜索空间时,一次只能处理一条约束上的一个值对;而借助ADD技术来压缩问题搜索空间,可以一次处理多条约束。算法首先通过01编码将CSP问题描述成伪布尔函数,并由ADD进行表示。然后基于传统弧一致性技术的算法思想,利用ADD的交、并和提取操作来实现约束传播和变量域过滤。最后将弧一致性的符号ADD算法嵌入到BT搜索算法中来实现对CSP的求解。对标准库中的测试用例以及随机生成的测试用例进行了实验仿真,结果表明,该算法求解CSP的时间既优于带弧一致性维护的回跳算法MAC3+BJ和MAC2001+BJ,也优于采用传统数据结构进行预处理的CSP求解算法BT+MPAC和BT+MPAC*。     

增强上下文的错误定位技术

错误定位就是寻找程序错误的位置.现有的错误定位方法大多利用测试用例的覆盖信息,以标识一组导致程序失效的可疑语句,却忽视了这些语句相互作用导致失效的上下文.因此,提出一种增强上下文的错误定位方法Context-FL,以构建上下文的方式来优化错误定位性能.Context-FL利用动态切片技术构建数据与控制相关性的错误传播上下文,显示了导致失效的语句之间传播依赖关系;然后,基于可疑值度量来区分上下文片段中不同语句的可疑度;最后,Context-FL以标记可疑值的上下文作为定位结果.实验结果表明,Context-FL优于8种典型错误定位方法.     

空间查询和路径搜索的集成处理策略

在数据库研究中,路径搜索和空间查询处理被认为是两个互不相关的领域,然而在处理具有空间约束的路径查询时,需要数据库系统提供路径计算和空间查询处理两方面的功能.为了处理路径计算中的空间约束,考虑了两类处理策略:(1) 空间运算是否在路径计算之前预处理;(2) 空间对象是否在路径计算之前预选取.基于这两类策略,应用现有的空间连接、R-树空间搜索和空间对象聚类技术,提出4种集成的空间路径查询处理方法.     

双局部粒子群算法解决环境经济调度问题

提出一种基于双局部最优的多目标粒子群优化算法,与可行解为优的约束处理方法相结合,来求解决非线性带约束的多目标电力系统环境经济调度问题。该算法针对传统多目标粒子群算法多样性低的局限性,通过对搜索空间的分割归类来增加帕累托最优解的多样性;并采用一种新的双局部最优来引导粒子的搜索,从而增强了算法的全局搜索能力。算法加入了可行解为优的约束处理方法对IEEE30节点六发电机电力系统环境经济负荷分配模型分别在几个不同复杂性问题的情况进行仿真测试,并与文献中的其他算法进行了比较。结果表明,改进的算法能够在保持帕累托最优解多样性的同时具有良好的收敛性能,更有效地解决电力系统环境经济调度问题。     

约束满足问题中一致性算法的分析与研究

约束满足问题是人工智能领域的重要研究方向,其求解方法有三种,搜索、一致性算法和约束传播,其中一致性算法通常通过缩减问题域来提高搜索算法的效率.着重介绍了几种常用的一致性算法,并对几种常用算法进行了分析、比较和研究.     

基于约束优化的Petri网可达性分析

Petri网的可达性判定问题是进行Petri网分析的基础。通过分析目前求解Petri网可达问题的判定方法和基于约束程序的Petri网可达问题判定方法,提出一种基于约束优化的Petri网可达问题判定方法,该方法是在状态方程法的基础上,利用约束程序寻求可行解,再利用优化求最优解,从而减少问题搜索的分支,达到减少状态方程的解空间的目的。最后通过实例的求解验证算法能够提高判定效率。     

DE+RBFNs based classification: A special attention to removal of inconsistency and irrelevant features

A novel approach for the classification of both balanced and imbalanced dataset is developed in this paper by integrating the best attributes of radial basis function networks and differential evolution. In addition, a special attention is given to handle the problem of inconsistency and removal of irrelevant features. Removing data inconsistency and inputting optimal and relevant set of features to a radial basis function network may greatly enhance the network efficiency (in terms of accuracy), at the same time compact its size. We use Bayesian statistics for making the dataset consistent, information gain theory (a kind of filter approach) for reducing the features, and differential evolution for tuning center, spread and bias of radial basis function networks. The proposed approach is validated with a few benchmarked highly skewed and balanced dataset retrieved from University of California, Irvine (UCI) repository. Our experimental result demonstrates promising classification accuracy, when data inconsistency and feature selection are considered to design this classifier.     

高光谱图像超分辨率重建技术研究进展

不同于传统图像（如灰度图像、RGB图像等）专注于保存目标场景的空间信息,高光谱图像蕴含丰富的空—谱信息,不仅可以保存目标的空间信息,还可以保存具有高可辨性的光谱信息。因此高光谱图像广泛应用于多种计算机视觉和遥感图像任务中,如目标检测、场景分类和目标追踪等。然而,在高光谱图像获取以及重建过程中仍然存在许多问题与瓶颈。如传统高光谱成像仪器在成像过程中通常会引入噪声,且获得的图像往往具有较低的空间分辨率,极大地影响了高光谱图像的质量,对后续数据分析任务造成了极大的困难。近年来,高光谱图像超分辨率重建技术研究得到了极大的发展,现有超分辨率重建方法可以大致分为两类,一类为空间超分辨率重建方法,可以通过直接提升高光谱图像的空间分辨率来获得高质量高光谱图像;另一类为光谱超分辨率重建方法,可以通过提升高空间分辨率图像的光谱分辨率来生成高质量高光谱图像。本文从高光谱图像超分辨率重建领域的新设计、新方法和应用场景出发,通过综合国内外前沿文献来梳理该领域的主要发展,重点论述高光谱图像超分辨率重建领域的发展现状、前沿动态、热点问题及趋势。     

Convergence in economic model predictive control with average constraints

In this paper, we thoroughly investigate various aspects of economic model predictive control with average constraints, i.e., constraints on average values of state and input variables. In particular, we first show that a certain time-varying output constraint has to be included into the MPC problem formulation in order to ensure fulfillment of these average constraints. Optimizing a general (possibly economic) performance criterion may result in a non-converging behavior of the corresponding closed-loop system. While such a behavior might be acceptable in some cases, it may be undesirable for other types of applications. Hence as a second contribution, we provide a Lyapunov-like analysis to conclude that indeed asymptotic convergence to the optimal steady-state follows if the system satisfies a certain dissipativity condition. Finally, for the case that this dissipativity property is not satisfied but still a convergent behavior of the closed-loop is required, we examine two different methods how convergence can be enforced within an economic MPC setup by imposing additional average constraints on the system. In the first method, an additional average constraint is defined which results in the system being dissipative, while the second consists of imposing an additional even zero-moment average constraint. We illustrate our results with various examples.     

稀疏正交普鲁克回归处理跨姿态人脸识别问题

正交普鲁克分析是 一种常用的处理矩阵近似问题的技术。最近,该技术被引入到正交普鲁克回归模型中来处理人脸姿态识别问题并取得了不错的效果。然而,这个模型对残差项使用了矩阵F范数约束,使得模型对于一些噪声(比如光照)非常敏感。为解决该问题,用更加鲁棒的1范约束替代原始的矩阵F范数约束,提出稀疏正交普鲁克回归模型。该模型可以由一个有效的交替迭代算法解决。在几个流行的人脸数据库上做了验证实验,实验结果证明该模型可以有效地处理人脸姿态变化。     

Efficient distributed low-cost backbone formation for wireless networks

Backbone has been used extensively in various aspects (e.g., routing, route maintenance, broadcast, scheduling) for wireless ad hoc or sensor networks recently. Previous methods are mostly designed to minimize the size of the backbone. However, in many applications, it is desirable to construct a backbone with small cost when each wireless node has a cost of being in the backbone. In this paper, we first show that previous methods specifically designed to minimize the backbone size may produce a backbone with large cost. Then, an efficient distributed method to construct a weighted backbone with low cost is proposed. We prove that the total cost of the constructed backbone is within a small constant factor of the optimum for homogeneous networks when either the nodes' costs are smooth (i.e., the maximum ratio of costs of adjacent nodes is bounded) or the network maximum node degree is bounded. We also show that, with a small modification, the backbone is efficient for unicast: the total cost (or hop) of the least cost (or hop) path connecting any two nodes using backbone is no more than three (or four) times the least cost (or hop) path in the original communication graph. Our theoretical. results are corroborated by our simulation studies. Finally, we discuss several possible ad hoc network applications of our proposed backbone formation algorithms.     

基于函数依赖与条件约束的数据修复方法

随着经济与信息技术的发展,在许多应用中均产生大量数据.然而,受硬件设备、人工操作、多源数据集成等诸多因素的影响,在这些应用之中往往存在较为严重的数据质量问题,特别是不一致性问题,从而无法有效管理数据.因此,首要的任务就是开发新型数据清洗技术来提升数据质量,以支持后续的数据管理与分析.现有工作主要研究基于函数依赖的数据修复技术,即以函数依赖来描述数据一致性约束,通过变更数据库中部分元组的属性值（而非增加/删除元组）来使得整个数据库遵循函数依赖集合.从一致性约束描述的角度来看,函数依赖并非是唯一的表达方式,还存在其他表达方式,例如硬约束、数量约束、等值约束、非等值约束等.然而,随着一致性约束种类的增加,其处理难度也远比仅有函数依赖的场景要困难.本文考虑以函数依赖与其他一致性约束共同表述数据库的一致性约束,并在此基础上设计数据修复算法,从而提升数据质量.实验结果表明,本文所提方法的执行效率较高.     

A multi-head neural network with unsymmetrical constraints for remaining useful life prediction

This paper proposes a multi-head neural network (MHNN) model with unsymmetrical constraints for remaining useful life (RUL) prediction of industrial equipment. Generally, the existing deep learning methods proposed for RUL prediction utilize symmetrical constraint loss functions such as the mean squared error function to calculate training errors. However, if the predicted RUL is much larger than the actual value in some safety–critical applications, severe damage may occur. To address this issue, an unsymmetrical constraint function is proposed as the loss function in this work that penalizes the late predictions (i.e., the predicted RUL is larger than the actual RUL) more strongly. In addition, an adjustable parameter is added to this function to adjust the model’s attention to the late predictions. In MHNN model, the bidirectional gated recurrent units (BGRU) and self-attention mechanism are employed to extract temporal features from the condition monitoring data. In addition, the structure of the multi-head neural network is adopted in the proposed model, helping to capture more degradation information by means of multiple identical and parallel networks. The proposed method is validated against a commonly used turbofan engine dataset. Compared with other latest methods on the same dataset, the proposed method is proven to be superior. Taking the FD004 dataset as an example, the score obtained by MHNN is 24.09% lower than that obtained by the best existing method.     

Effects of stability on model composition effort: an exploratory study

Model composition plays a central role in many software engineering activities, e.g., evolving design models to add new features. To support these activities, developers usually rely on model composition heuristics. The problem is that the models to-be-composed usually conflict with each other in several ways and such composition heuristics might be unable to properly deal with all emerging conflicts. Hence, the composed model may bear some syntactic and semantic inconsistencies that should be resolved. As a result, the production of the intended model is an error-prone and effort-consuming task. It is often the case that developers end up examining all parts of the output composed model instead of prioritizing the most critical ones, i.e., those that are likely to be inconsistent with the intended model. Unfortunately, little is known about indicators that help developers (1) to identify which model is more likely to exhibit inconsistencies, and (2) to understand which composed models require more effort to be invested. It is often claimed that software systems remaining stable over time tends to have a lower number of defects and require less effort to be fixed than unstable systems. However, little is known about the effects of software stability in the context of model evolution when supported by composition heuristics. This paper, therefore, presents an exploratory study analyzing stability as an indicator of inconsistency rate and resolution effort on model composition activities. Our findings are derived from 180 compositions performed to evolve design models of three software product lines. Our initial results, supported by statistical tests, also indicate which types of changes led to lower inconsistency rate and lower resolution effort.     

面向多语料库的通用事件指代消解

事件同指消解对篇章理解、信息抽取意义重大。该文在事件抽取完成的基础上聚焦事件同指,给出了一个基于卷积神经网络的事件同指消解完整框架,针对实例分布不均衡问题给出了基于事件语义类别和时态信息的事件兼容性过滤策略。为了最大化适用不同的事件标注策略,提出了最小事件本身描述和事件间关系描述相结合的特征表示方法。针对基于事件对模型进行同指消解产生的局部最优问题,给出了一种全局优化的后处理方案。在KBP2015和ACE2005语料上的各类实验表明,上述三个解决方案均能有效解决问题,提升整个事件同指消解平台的性能。