期刊文献+
共找到3篇文章
< 1 >
每页显示 20 50 100
融合关联矩阵自学习和显式秩约束的数据表示分簇算法 被引量:2
1
作者 郑建炜 鞠振宇 +1 位作者 朱文博 王万良 《计算机学报》 EI CSCD 北大核心 2019年第3期497-512,共16页
复杂异构分布的高维数据在大数据时代随处可见,高效地挖掘其子空间结构并进行准确的分簇是机器视觉和模式识别领域的研究热点.低秩表示算法(Low-Rank Representation,LRR)因其优越的低维子空间挖掘能力而备受关注,其性能很大程度上取决... 复杂异构分布的高维数据在大数据时代随处可见,高效地挖掘其子空间结构并进行准确的分簇是机器视觉和模式识别领域的研究热点.低秩表示算法(Low-Rank Representation,LRR)因其优越的低维子空间挖掘能力而备受关注,其性能很大程度上取决于关联矩阵的构建,常见的方法都是通过原始输入数据或表示系数直接一次成形.然而,这些方法都采用独立的步骤进行表示系数计算以及关联矩阵学习,无法保证总体算法的最优性.针对该问题,该文提出一种新的LRR型数据表示分簇法(Data Representation Clustering,DRC)应用于实际子空间分割问题.首先,为实现模型的快速求解,DRC保留了基本数据表示框架中的光滑正则项并剔除了非负性、稀疏性等复杂约束;其次,将相似度矩阵的自适应学习策略添加至统一的数据表示框架,联合原始输入数据和表示系数确保目标关联矩阵在无噪环境下具备明确的对角分布结构.最后,对关联矩阵对应的Laplacian矩阵添加一种新的秩约束,在含噪环境下引导相似度连接结构与簇目标数的一致性.采用交替更新法对模型进行求解,保证目标函数单变量优化的全局最优性以及整体收敛性.人工合成数据和8个公开数据集的实验结果表明,DRC算法在分簇精度、归一化互信息、参数敏感性等指标上都具有优秀的性能. 展开更多
关键词 关联矩阵 低秩表示 谱分簇 拉普拉斯正则项 归一化互信息
下载PDF
A cooperative spectrum sensing results transmission scheme with LT code based on energy efficiency priority
2
作者 Song Xiaoqin Jin Hui +2 位作者 Tan Yazhu Hu Jing Song Tiecheng 《Journal of Southeast University(English Edition)》 EI CAS 2018年第4期444-450,共7页
To deal with a sharp increase in transmission energy consumption due to the presence of a large number of secondary users( SUs),an energy-efficient cooperative spectrum sensing results transmission scheme is proposed ... To deal with a sharp increase in transmission energy consumption due to the presence of a large number of secondary users( SUs),an energy-efficient cooperative spectrum sensing results transmission scheme is proposed for cognitive radio systems. First,a cluster-based structure is introduced into the sensing results transmission scheme to reduce the transmission power. Then,the centralized sensing results transmission model is employed,and the non-fixed code rate Luby transform( LT)code is selected as the channel coding since its code rate can dynamically adapt to channel conditions and therefore avoid unnecessary redundancy in the transmission power.Furthermore,an improved optimal degree distribution( ODD)is designed for the LT code. The simulation results show that the choice of the appropriate parameters in degree distribution is very helpful for the LT code to achieve a promising performance. The ODD with optimized parameters can achieve more than 2 d B performance gain than other typical degree distributions when the bit error rate( BER) is 10-3. The energy consumption of the proposed scheme is not only at least71. 4% lower than that of the non-coding system,but also lower than that of the convolutional coding system with different code rates. Meanwhile, the energy consumption can be further reduced in the case that a suitable clustering method is selected. 展开更多
关键词 cooperative spectrum sensing ENERGY-EFFICIENT cluster-based Luby transform (LT) code degree distribution
下载PDF
Kinetically controlled, high-yield, direct synthesis of [Au_(25)(SePh)_(18)]^-TOA^+
3
作者 SONG YongBo CAO TianTian +3 位作者 DENG HuiJuan ZHU XiuYi LI Peng ZHU ManZhou 《Science China Chemistry》 SCIE EI CAS 2014年第9期1218-1224,共7页
In this article, we present a facile, direct, synthetic approach of preparing monodisperse [Au2s(SePh)ls]- nanoclusters in high yield. In this synthetic approach, two-phase Brust-Schiffrin method is used. Both PhSeH... In this article, we present a facile, direct, synthetic approach of preparing monodisperse [Au2s(SePh)ls]- nanoclusters in high yield. In this synthetic approach, two-phase Brust-Schiffrin method is used. Both PhSeH and NaBH4 should be added drop-wise to the solution of Au (III) at the same time. The formula and molecular purity of [Au25(SePh)ls] TOA+ clusters are characterized by MALDI-TOF mass spectrometry, NMR and TGA analysis. Furthermore, some critical parameters to obtain pure [Au25(SePh)18]-TOA+ are identified, including the NaBH4-to-Au ratio, the selenolate-to-Au ratio and the temperature. The facile, direct, high yield synthetic method can be widely applied in the theoretical research of Au clusters protected by selenol. 展开更多
关键词 selenophenol Au cluster Au2s(SePh)18 kinetically controlled atomic precision
原文传递
上一页 1 下一页 到第
使用帮助 返回顶部