目的:探讨AGFA COMPACT PLUS CR数字转换器的维修经验和体会。方法:通过掌握的维修实例了解设备的常见故障和维修方式。结果:及时有效地解决常见故障。结论:通过对AGFA COMPACT PLUS CR数字转换器的故障分析,学习掌握对设备故障的判断...目的:探讨AGFA COMPACT PLUS CR数字转换器的维修经验和体会。方法:通过掌握的维修实例了解设备的常见故障和维修方式。结果:及时有效地解决常见故障。结论:通过对AGFA COMPACT PLUS CR数字转换器的故障分析,学习掌握对设备故障的判断、分析及维修并能对设备进行保养。展开更多
In this paper, a novel approach for service substitutions based on the service type in terms of its interface type and behavior semantics is proposed. In order to analyze and verify behavior-consistent service substit...In this paper, a novel approach for service substitutions based on the service type in terms of its interface type and behavior semantics is proposed. In order to analyze and verify behavior-consistent service substitutions in dynamic environments, we first present a formal language to describe services from control-flow perspective, then introduce a type and effect system to infer conservative approximations of all possible behaviors of these services. The service behaviors are represented by concurrent behavior expressions （CBEs）. Built upon the interpretation of CBEs, behavior-consistent service substitutions are defined and analyzed by subtyping technology. The correctness of the analysis approach is guaranteed by type safety theorem, which is mechanically proved in the Coq proof assistant. Finally, applications in web services show that our method is effective and feasible.展开更多
The present paper deals with the existence of periodic orbits in the Circular Restricted Four-Body Problem (CR4BP) in two-dimensional co-ordinate system when the second primary is a triaxial rigid body and the third p...The present paper deals with the existence of periodic orbits in the Circular Restricted Four-Body Problem (CR4BP) in two-dimensional co-ordinate system when the second primary is a triaxial rigid body and the third primary of inferior mass (in comparison of the other primaries) is placed at triangular libration point L4 of the Circular Restricted Three-Body Problem (CR3BP). With the help of generating solutions, we formed a basis for the existence of periodic orbits, then an analytical approach given by Hassan et al. [1], was applied to our model of equilateral triangular configuration. It is found that in general solution also;the character of periodic orbits is conserved. For verification of the existence of periodic orbits, we have applied the criterion of Duboshin [2] and found satisfied.展开更多
建筑物的变化检测在遥感图像处理和模式识别领域中具有重要意义,但在深度学习算法的应用中,数据标注一直是一个显著的挑战,特别是在变化检测的场景下。因此,针对基于深度学习的变化检测算法中数据标注的难题,本文提出了一种半监督学习...建筑物的变化检测在遥感图像处理和模式识别领域中具有重要意义,但在深度学习算法的应用中,数据标注一直是一个显著的挑战,特别是在变化检测的场景下。因此,针对基于深度学习的变化检测算法中数据标注的难题,本文提出了一种半监督学习方法。该方法采用了融合双时相特征的孪生网络来进行特征提取,并构建了一个教师-学生网络框架以实施模型的半监督训练。为了进一步提升半监督变化检测的准确度,本文在深度特征上引入了随机扰动,以此来实现一致性正则化。此外,在图像深度特征的层面上,本文还提出通过捕获双时相图像特征的差异来形成决策边界以区分双时相图变化的方法。该方法在Levir-CD和WHU_Building两个公开数据集上分别实现了83.04%和85.57%的交并比(Intersection over Union,IoU)。实验结果表明,本文提出的方法在使用少量标注数据的前提下,能够达到与全监督训练相近的性能。展开更多
现有的半监督声音事件检测方法直接使用强标签合成样本、弱标签真实样本和无标签真实样本进行训练,以缓解标签样本量不足的问题。然而,合成和真实数据域之间存在不可避免的分布差异,这种差异会干扰模型梯度优化方向,从而限制模型的泛化...现有的半监督声音事件检测方法直接使用强标签合成样本、弱标签真实样本和无标签真实样本进行训练,以缓解标签样本量不足的问题。然而,合成和真实数据域之间存在不可避免的分布差异,这种差异会干扰模型梯度优化方向,从而限制模型的泛化能力。针对这一问题,基于元学习(Meta Learning)提出了一种新颖的半监督声音事件检测学习范式MMT(Meta Mean Teacher)。具体来说,对于每个训练批次的数据,将其分为由合成样本组成的元训练集和由真实样本组成的元测试集;将模型在元训练集上计算的元梯度作为元测试梯度更新的指导,使模型感知并学习到更具泛化性的知识。在DCASE2021任务4数据集的测试集上进行对比实验,结果表明,相较于官方基线,所提出的学习范式MMT在F1,PSDS1和PSDS2指标上分别提升了8.9%,6.6%和1.1%;相较于当前的先进方法,所提出的学习范式MMT同样表现出了显著的性能优势。展开更多
基金the National Natural Science Foundation of China(Nos.61232007 and 91118004)the Innovation Program of Shanghai Municipal Education Commission(No.13ZZ023)
文摘In this paper, a novel approach for service substitutions based on the service type in terms of its interface type and behavior semantics is proposed. In order to analyze and verify behavior-consistent service substitutions in dynamic environments, we first present a formal language to describe services from control-flow perspective, then introduce a type and effect system to infer conservative approximations of all possible behaviors of these services. The service behaviors are represented by concurrent behavior expressions （CBEs）. Built upon the interpretation of CBEs, behavior-consistent service substitutions are defined and analyzed by subtyping technology. The correctness of the analysis approach is guaranteed by type safety theorem, which is mechanically proved in the Coq proof assistant. Finally, applications in web services show that our method is effective and feasible.
文摘The present paper deals with the existence of periodic orbits in the Circular Restricted Four-Body Problem (CR4BP) in two-dimensional co-ordinate system when the second primary is a triaxial rigid body and the third primary of inferior mass (in comparison of the other primaries) is placed at triangular libration point L4 of the Circular Restricted Three-Body Problem (CR3BP). With the help of generating solutions, we formed a basis for the existence of periodic orbits, then an analytical approach given by Hassan et al. [1], was applied to our model of equilateral triangular configuration. It is found that in general solution also;the character of periodic orbits is conserved. For verification of the existence of periodic orbits, we have applied the criterion of Duboshin [2] and found satisfied.
文摘建筑物的变化检测在遥感图像处理和模式识别领域中具有重要意义,但在深度学习算法的应用中,数据标注一直是一个显著的挑战,特别是在变化检测的场景下。因此,针对基于深度学习的变化检测算法中数据标注的难题,本文提出了一种半监督学习方法。该方法采用了融合双时相特征的孪生网络来进行特征提取,并构建了一个教师-学生网络框架以实施模型的半监督训练。为了进一步提升半监督变化检测的准确度,本文在深度特征上引入了随机扰动,以此来实现一致性正则化。此外,在图像深度特征的层面上,本文还提出通过捕获双时相图像特征的差异来形成决策边界以区分双时相图变化的方法。该方法在Levir-CD和WHU_Building两个公开数据集上分别实现了83.04%和85.57%的交并比(Intersection over Union,IoU)。实验结果表明,本文提出的方法在使用少量标注数据的前提下,能够达到与全监督训练相近的性能。
文摘现有的半监督声音事件检测方法直接使用强标签合成样本、弱标签真实样本和无标签真实样本进行训练,以缓解标签样本量不足的问题。然而,合成和真实数据域之间存在不可避免的分布差异,这种差异会干扰模型梯度优化方向,从而限制模型的泛化能力。针对这一问题,基于元学习(Meta Learning)提出了一种新颖的半监督声音事件检测学习范式MMT(Meta Mean Teacher)。具体来说,对于每个训练批次的数据,将其分为由合成样本组成的元训练集和由真实样本组成的元测试集;将模型在元训练集上计算的元梯度作为元测试梯度更新的指导,使模型感知并学习到更具泛化性的知识。在DCASE2021任务4数据集的测试集上进行对比实验,结果表明,相较于官方基线,所提出的学习范式MMT在F1,PSDS1和PSDS2指标上分别提升了8.9%,6.6%和1.1%;相较于当前的先进方法,所提出的学习范式MMT同样表现出了显著的性能优势。
