基于生成对抗网络的细胞形变动态分类

提出了一种基于生成对抗网络的细胞形变动态分类方法,以活细胞视频中的细胞形变动态为对象,引入分类器辅助的生成对抗网络结构同步训练生成对抗网络和分类网络,通过生成对抗网络产生的数据提高了原本分类网络分辨细胞形变动态的性能.首先,细胞动态图像被用于将活细胞视频中的时间维度进行压缩,使其从视频域映射到图像域以方便生成对抗网络的构建.其次,基于分类器辅助的生成对抗网络结构,将分类网络的分类信息作为辅助信息来改善生成对抗网络对多类样本的生成,同时生成网络生成的多类样本可以反过来优化分类网络对于细胞动态形变的分类性能.在构建的活细胞视频数据库上,可以验证提出方法能有效地捕获细胞视频中的空时细胞形变动态,并且其分类的性能优于其它主流方法.     

无线传感器网络多目标路由的改进蚁群算法

提出一种集能耗、时延、鲁棒性和传输效率等于一体的多目标路由,且各目标的权重可以根据实际情况进行调节,具有较强的灵活性,提出一种正反馈和负反馈并存机制的蚁群算法,其主要思想是,若前路径比以往求得的最好路径性能更优,则当前路径信息素将加强,同时用当前路径取代最好路径,否则当前路径信息素减弱.用该改进的蚁群算法求解无线传感器网络多目标路由问题,实验数据表明:改进的蚁群算法的路由各方面性能良好,并优于目前其他典型路由.     

二维具时滞Hopfield神经网络动力学行为研究

Hopfield神经网络模型是对人脑神经网络的某些机理与机制的抽象,它可以模拟大脑实现联想记忆和优化计算等方面的功能.本文对一类二维具时滞的Hopfield神经网络模型进行了理论分析,给出了系统可能出现的稳定平衡点、周期解以及混沌奇异吸引子的参数选取信息,并在理论分析的基础上进行了数值模拟,用计算机描绘出解曲线图.本结果对人工神经网络的应用提供参数依据.     

Online Distributed Fault Detection of Sensor Measurements

In wireless sensor networks (WSNs), a faulty sensor may produce incorrect data and transmit them to the other sensors. This would consume the limited energy and bandwidth of WSNs. Furthermore, the base station may make inappropriate decisions when it receives the incorrect data sent by the faulty sensors. To solve these problems, this paper develops an online distributed algorithm to detect such faults by exploring the weighted majority vote scheme. Considering the spatial correlations in WSNs, a faulty sensor can diagnose itself through utilizing the spatial and time information provided by its neighbor sensors. Simulation results show that even when as many as 30% of the sensors are faulty, over 95% of faults can be correctly detected with our algorithm. These results indicate that the proposed algorithm has excellent performance in detecting fault of sensor measurements in WSNs.     

传感器网络空间关键字近似Top-k查询算法

提出了一种基于节点剪枝的Top-k査询算法.定义集合支配区域与剪枝规则,根据用户查询位置的变动,按照给定的剪切规则,对传感器网络节点进行剪枝处理,将符合规则的节点预先筛除,运用统计学抽样理论,设计了空间关键字近似Top-k查询算法.在2个真实数据集上进行验证.实验结果表明,该算法能够满足传感器网络环境中用户的需求,并且相比于边界距离增量算法(IBD)与基于改进的MW-Voronoi区域的空间关键字查询算法(MSK-uvr),其网络通信开销较低.     

On Peer-Assisted Data Dissemination in Data Center Networks: Analysis and Implementation简

Data Center Networks （DCNs） are the fundamental infrastructure for cloud computing. Driven by the massive parallel computing tasks in cloud computing, one-to-many data dissemination becomes one of the most important traffic patterns in DCNs. Many architectures and protocols are proposed to meet this demand. However, these proposals either require complicated configurations on switches and servers, or cannot deliver an optimal performance. In this paper, we propose the peer-assisted data dissemination for DCNs. This approach utilizes the rich physical connections with high bandwidths and mutli-path connections, to facilitate efficient one-to-many data dissemination. We prove that an optimal P2P data dissemination schedule exists for FatTree, a specially- designed DCN architecture. We then present a theoretical analysis of this algorithm in the general multi-rooted tree topology, a widely-used DCN architecture. Additionally, we explore the performance of an intuitive line structure for data dissemination. Our analysis and experimental results prove that this simple structure is able to produce a comparable performance to the optimal algorithm. Since DCN applications heavily rely on virtualization to achieve optimal resource sharing, we present a general implementation method for the proposed algorithms, which aims to mitigate the impact of the potentially-high churn rate of the virtual machines.     

汽轮机故障诊断中的信息熵融合

在汽轮机轴系振动故障模拟试验的基础上,对大量故障模拟试验数据进行计算,建立了典型故障的4种信息熵样本.采用概率神经网络对故障信号的4种信息熵特征进行融合研究,并将融合结果与最小距离分类器的分类效果进行了对照分析.研究表明,概率神经网络可实现对训练样本100%的正确识别率,对"陌生"样本的正确识别率也超过80%,其识别效果远远超过最小距离分类器.可见,概率神经网络综合了贝叶斯分类器和神经网络的优势,在汽轮机故障模式分类方面具有明显的优势,利用概率神经网络融合信号的信息熵特征实现汽轮机轴系故障模式识别是一种可行有效的方法.     

Reliable and Energy Efficient Target Coverage for Wireless Sensor Networks

A critical aspect of applications with Wireless Sensor Networks (WSNs) is network lifetime.Power-constrained WSNs are usable as long as they can communicate sense data to a processing node.Poor communication links and hazardous environments make the WSNs unreliable.Existing schemes assume that the state of a sensor covering targets is binary:success (covers the targets) or failure (cannot cover the targets).However,in real WSNs,a sensor covers targets with a certain probability.To improve WSNs’ reliability,we should consider that a sensor covers targets with users’ satisfied probability.To solve this problem,this paper first introduces a failure probability into the target coverage problem to improve and control the system reliability.Furthermore,we model the solution as the α-Reliable Maximum Sensor Covers (α-RMSC) problem and design a heuristic greedy algorithm that efficiently computes the maximal number of α-Reliable sensor covers.To efficiently extend the WSNs lifetime with users’ pre-defined failure probability requirements,only the sensors from the current active sensor cover are responsible for monitoring all targets,while all other sensors are in a low-energy sleep mode.Simulation results validate the performance of this algorithm,in which users can precisely control the system reliability without sacrificing much energy consumption.     

铜合金时效性能的神经网络预测模型

运用BP神经网络算法，对时效试验数据进行训练，建立了Cu-0．30Cr-0．15Zr合金时效后硬度和导电性与时效时间和时效温度的映射模型，从而可预测铜合金在一定时效条件下的硬度和导电性，预测结果与实测值吻合较好，表明神经网络用于铜合金的时效性能预测是可行的，预测的准确性取决于用于训练网络的铜合金时效试验数据的数量和质量。     

弓型折流板换热器动态响应神经网络预测

采用ARX动态模型设计了一种用于换热器动态特性辨识及预测的人工神经网络结构模型,基于有限的实验数据将人工神经网络技术应用于以水、油为换热工质的弓型折流板换热器动态特性的预测当中.采用Levenberg-Marquardt算法对网络进行训练和测试,辨识和预测了换热器单侧流量扰动情况下油侧出口温度的响应情况及换热器油侧进口发生温度扰动情况下油侧出口温度响应情况,并与数值计算进行了对比,神经网络的预测结果好于数值计算.分析了神经网络模型的泛化能力.计算结果表明,人工神经网络对于复杂系统的动态辨识及预测是相当成功的,与基于数理模型的数值预测相比,有更好的预测精度,且模型的泛化能力也很强.