Cooperative spectrum monitoring with multiple sensors has been deemed as an efficient mechanism for improving the monitoring accuracy and enlarging the monitoring area in wireless sensor networks.However,there exists ...Cooperative spectrum monitoring with multiple sensors has been deemed as an efficient mechanism for improving the monitoring accuracy and enlarging the monitoring area in wireless sensor networks.However,there exists redundancy among the spectrum data collected by a sensor node within a data collection period,which may reduce the data uploading efficiency.In this paper,we investigate the inter-data commonality detection which describes how much two data have in common.We define common segment set and divide it into six categories firstly,then a method to measure a common segment set is conducted by extracting commonality between two files.Moreover,the existing algorithms fail in finding a good common segment set,so Common Data Measurement(CDM)algorithm that can identify a good common segment set based on inter-data commonality detection is proposed.Theoretical analysis proves that CDM algorithm achieves a good measurement for the commonality between two strings.In addition,we conduct an synthetic dataset which are produced randomly.Numerical results shows that CDM algorithm can get better performance in measuring commonality between two binary files compared with Greedy-String-Tiling(GST)algorithm and simple greedy algorithm.展开更多
传统的多模态过程故障等级评估方法对模态之间的共性特征考虑较少,导致当被评估模态故障信息不充分时,评估的准确性较低.针对此问题,首先,提出一种共性–个性深度置信网络(Common and specific deep belief network,CS-DBN),该网络充分...传统的多模态过程故障等级评估方法对模态之间的共性特征考虑较少,导致当被评估模态故障信息不充分时,评估的准确性较低.针对此问题,首先,提出一种共性–个性深度置信网络(Common and specific deep belief network,CS-DBN),该网络充分利用深度置信网络(Deep belief network,DBN)的深度分层特征提取能力,通过度量多模态数据间分布的相似性和差异性,进一步得到能够反映多模态过程共有信息的共性特征以及反映每个模态独有信息的个性特征;其次,基于CS-DBN,利用多模态过程的已知故障等级数据生成多模态共性–个性特征集,通过加权逻辑回归构建故障等级评估模型;最后,将所提方法应用于带钢热连轧生产过程的故障等级评估中.应用结果表明,随着多模态故障等级数据的增加,所提方法的评估准确率逐渐增加,当故障信息充足时,评估准确率可达98.75%;故障信息不足时,与传统方法相比,评估准确率提升近10%.展开更多
为提高运动想象脑机接口识别准确率,结合共空间模式(common spatial pattern,CSP)和卷积神经网络(convolutional neural network,CNN)方法,提出一种改进滤波器组共空间模式(filter bank common spatial pattern,FBCSP)和CNN的算法,用于...为提高运动想象脑机接口识别准确率,结合共空间模式(common spatial pattern,CSP)和卷积神经网络(convolutional neural network,CNN)方法,提出一种改进滤波器组共空间模式(filter bank common spatial pattern,FBCSP)和CNN的算法,用于多分类运动想象脑电信号识别任务。信号预处理后,使用包含重叠频带的FBCSP计算空间投影矩阵,数据经过投影得到更有区分度的特征序列。然后将特征序列以二维排列方式输入搭建的CNN模型中进行分类。所提出方法在脑机接口竞赛数据集2a和Ⅲa上验证,并和其他文献方法对比。结果表明,本文方法一定程度上提高了运动想象脑电信号的分类准确率,为运动想象研究提供了一个有效办法。展开更多
基金Supported by National High Technology Research and Development Program of China （863 Program） （2012AA06A404）, National Natural Science Foundation of China （61004074, 61134001, 21076179）, National Key Technology Support Program of China （2009BAG12A 08）, and Fundamental Research Funds for the Central Universities （2010QNA5001）
基金supported in part by the National Natural Science Foundation of China(No.61901328)the China Postdoctoral Science Foundation (No. 2019M653558)+1 种基金the Fundamental Research Funds for the Central Universities (No. CJT150101)the Key project of National Natural Science Foundation of China (No. 61631015)
文摘Cooperative spectrum monitoring with multiple sensors has been deemed as an efficient mechanism for improving the monitoring accuracy and enlarging the monitoring area in wireless sensor networks.However,there exists redundancy among the spectrum data collected by a sensor node within a data collection period,which may reduce the data uploading efficiency.In this paper,we investigate the inter-data commonality detection which describes how much two data have in common.We define common segment set and divide it into six categories firstly,then a method to measure a common segment set is conducted by extracting commonality between two files.Moreover,the existing algorithms fail in finding a good common segment set,so Common Data Measurement(CDM)algorithm that can identify a good common segment set based on inter-data commonality detection is proposed.Theoretical analysis proves that CDM algorithm achieves a good measurement for the commonality between two strings.In addition,we conduct an synthetic dataset which are produced randomly.Numerical results shows that CDM algorithm can get better performance in measuring commonality between two binary files compared with Greedy-String-Tiling(GST)algorithm and simple greedy algorithm.
文摘传统的多模态过程故障等级评估方法对模态之间的共性特征考虑较少,导致当被评估模态故障信息不充分时,评估的准确性较低.针对此问题,首先,提出一种共性–个性深度置信网络(Common and specific deep belief network,CS-DBN),该网络充分利用深度置信网络(Deep belief network,DBN)的深度分层特征提取能力,通过度量多模态数据间分布的相似性和差异性,进一步得到能够反映多模态过程共有信息的共性特征以及反映每个模态独有信息的个性特征;其次,基于CS-DBN,利用多模态过程的已知故障等级数据生成多模态共性–个性特征集,通过加权逻辑回归构建故障等级评估模型;最后,将所提方法应用于带钢热连轧生产过程的故障等级评估中.应用结果表明,随着多模态故障等级数据的增加,所提方法的评估准确率逐渐增加,当故障信息充足时,评估准确率可达98.75%;故障信息不足时,与传统方法相比,评估准确率提升近10%.
文摘为提高运动想象脑机接口识别准确率,结合共空间模式(common spatial pattern,CSP)和卷积神经网络(convolutional neural network,CNN)方法,提出一种改进滤波器组共空间模式(filter bank common spatial pattern,FBCSP)和CNN的算法,用于多分类运动想象脑电信号识别任务。信号预处理后,使用包含重叠频带的FBCSP计算空间投影矩阵,数据经过投影得到更有区分度的特征序列。然后将特征序列以二维排列方式输入搭建的CNN模型中进行分类。所提出方法在脑机接口竞赛数据集2a和Ⅲa上验证,并和其他文献方法对比。结果表明,本文方法一定程度上提高了运动想象脑电信号的分类准确率,为运动想象研究提供了一个有效办法。
