基于粗糙集的模糊神经网络降水预报模型研究

为了提高模糊神经网络预测模型的预测能力,提出了基于粗集理论方法选择自变量作为模型输入的模糊神经网络预报建模方法.以短期降水预报作为研究对象,利用粗集理论的条件属性约简计算分析方法,对初选得到的预报因子矩阵进行属性约简,剔除不相关的属性,找出与预报量直接相关的预报因子,建立模糊神经网络的降水预报模型.实际的预报试验结果表明,该预报方法的预报精度明显高于由逐步回归方法选择预报因子作为模型输入的模糊神经网络预报模型及中国气象局T213数值预报模式的预报结果.     

梯度均值预测滤波算法及其应用

测量误差不可避免的存在于各种测量系统之中.针对缓慢变化的测量数据中存在的随机误差,结合传统滤波方法的优点,提出了梯度均值预测滤波算法,阐述了算法的基本思想和流程,并结合桥梁挠度连续测量系统的测量数据进行了滤波实验,结果表明,该算法能够有效去除缓慢变化的测量数据中的干扰信号,滤波效果明显.     

大气湍流图像像素偏差机理分析

对大气湍流导致图像畸变和像素偏差的因素进行了深入分析,对大气湍流状态下激光发射远距离光束的光束漂移均方值进行仿真,分析光束漂移均方值曲线,为后续对图像畸变、像素偏差校正及克服湍流影响提供有效的理论依据。     

自适应带宽均值移动算法及目标跟踪

首先提出了一种经典均值移动算法的推广算法,即自适应带宽均值移动算法,进而提出了基于自适应带宽均值移动的二维视频目标跟踪算法（ABMSOT）.前者提出了在带宽自适应情况下均值移动算法求取局部极值的框架步骤,后者可实时跟踪目标的位置、大小和方向.在ABMSOT算法中,目标模型和候选模型采用自适应带宽核函数加权特征直方图描述,目标模型和候选模型的相似性采用Bhattacharyya系数度量;通过迭代两步法搜索到目标最有可能的位置、大小和方向.第一步执行一次均值移动迭代搜索目标位置,第二步计算出最能描述目标区域大小和方向的带宽矩阵.从理论上证明了两个算法的收敛性,并通过实验证明了ABMSOT算法能实时跟踪目标的位置、大小和方向.     

A new scaling algorithm for the maximum mean cut problem

We present a new scaling algorithm for the maximum mean cut problem. The mean of a cut is defined by the cut capacity divided by the number of arcs crossing the cut. The algorithm uses an approximate binary search and solves the circulation feasibility problem with relaxed capacity bounds. The maximum mean cut problem has recently been studied as a dual analogue of the minimum mean cycle problem in the framework of the minimum cost flow problem by Ervolina and McCormick. A networkN=(G, lower, upper) with lower and upper arc capacities is said to be -feasible ifN has a feasible circulation when we relax the capacity bounds by ; that is, we use (lower(a)- , upper(a)+) bounds instead of (lower(a), upper(a)) bounds for each arca A. During an approximate binary search we maintain two bounds,LB andUB, such thatN is LB-infeasible andUB-feasible, and we reduce the interval size (LB, UB) by at least one-third at each iteration. For a graph withn vertices, m arcs, and integer capacities bounded byU, the running time of this algorithm is O(mn log(nU). This time bound is better than the time achieved by McCormick and Ervolina under thesimilarity condition (that is,U=O(n^o(1))). Our algorithm can be naturally used for the circulation feasibility problem, and thus provides a new scaling algorithm for the minimum cut problem.Research supported by a grant-in-aid of the Ministry of Education, Science and Culture of Japan.     

基于MELMD-ICA的光纤振动信号降噪方法

针对分布式光纤传感系统所采集含噪信号,提出一种改进集成局部均值分解(MELMD)联合独立成分分析(ICA)的降噪方法,引入排列熵判决机制提高抑制模态混叠与虚假分量能力。首先使用MELMD方法分解含噪信号得到乘积函数(PF)并进行信号重构;将含噪信号和重构信号求差得到虚拟噪声,构造虚拟通道;然后使用ICA对含噪信号和虚拟通道进行信噪分离,得到最终结果。通过实验验证,该方法与EMD-ICA,EEMD-ICA,MELMD相比,能更好地消除信号中的噪声,保留信号的特征信息。     

用于高能物理实验电子读出芯片的低噪声锁相环芯片设计

基于TSMC 180 nm工艺设计并流片测试了一款用于高能物理实验的电子读出系统的低噪声、低功耗锁相环芯片。该芯片主要由鉴频鉴相器、电荷泵、环路滤波器、压控振荡器和分频器等子模块组成,在锁相环电荷泵模块中,使用共源共栅电流镜结构精准镜像电流以减小电流失配和用运放钳位电压进一步减小相位噪声。测试结果表明,该锁相环芯片在1.8 V电源电压、输入50 MHz参考时钟条件下,可稳定输出200 MHz的差分时钟信号,时钟均方根抖动为2.26 ps(0.45 mUI),相位噪声在1 MHz频偏处为-105.83 dBc/Hz。芯片整体功耗实测为23.4 mW,锁相环核心功耗为2.02 mW。     

基于黎曼流形的多视角谱聚类算法

现有的多视角谱聚类算法大多只线性结合了各视角的基拉普拉斯矩阵,未考虑不同视角数据的差异性对最优拉普拉斯矩阵的影响,存在聚类性能受限的问题。提出一种基于黎曼几何均值与高阶拉普拉斯矩阵的谱聚类算法(RMMSC),挖掘多视角数据中的高阶连接信息与流形信息,提高最优拉普拉斯矩阵对各视角的信息利用率。按一定的权重线性结合数据单一视角的各阶拉普拉斯矩阵,得到每个视角的基拉普拉斯矩阵,通过低阶与高阶连接信息的结合使用,充分体现多视角数据集的全局结构。在此基础上,计算各视角基拉普拉斯矩阵的黎曼几何均值,将其作为最优拉普拉斯矩阵输入谱聚类算法,得到聚类结果。相比于传统矩阵算数均值的计算,基于黎曼流形的黎曼几何均值能够更好地恢复互补层数据的流形信息。实验结果表明,RMMSC在多组标准数据集上聚类效果优于ONMSC、MLAN、AMGL等算法。其中,在Flower17数据集上,精确度较基准算法ONMSC提高了2.14%,纯度提高了1.7%,且收敛性较好。     

Implicit Continuous User Authentication for Mobile Devices based on Deep Reinforcement Learning

The predominant method for smart phone accessing is confined to methods directing the authentication by means of Point-of-Entry that heavily depend on physiological biometrics like, fingerprint or face. Implicit continuous authentication initiating to be loftier to conventional authentication mechanisms by continuously confirming users’ identities on continuing basis and mark the instant at which an illegitimate hacker grasps dominance of the session. However, divergent issues remain unaddressed. This research aims to investigate the power of Deep Reinforcement Learning technique to implicit continuous authentication for mobile devices using a method called, Gaussian Weighted Cauchy Kriging-based Continuous Czekanowski’s (GWCK-CC). First, a Gaussian Weighted Non-local Mean Filter Preprocessing model is applied for reducing the noise present in the raw input face images. Cauchy Kriging Regression function is employed to reduce the dimensionality. Finally, Continuous Czekanowski’s Classification is utilized for proficient classification between the genuine user and attacker. By this way, the proposed GWCK-CC method achieves accurate authentication with minimum error rate and time. Experimental assessment of the proposed GWCK-CC method and existing methods are carried out with different factors by using UMDAA-02 Face Dataset. The results confirm that the proposed GWCK-CC method enhances authentication accuracy, by 9%, reduces the authentication time, and error rate by 44%, and 43% as compared to the existing methods.