在大数据时代,数据的类别标签数量激增,对现有的分类方法构成了重大挑战。为解决此问题,利用类别间的相似性,将数据类别标签以层次化方式处理。但现有的类别间相似性度量均使用欧氏距离,由于欧氏距离无法有效处理高维数据,因此,受Tanim...在大数据时代,数据的类别标签数量激增,对现有的分类方法构成了重大挑战。为解决此问题,利用类别间的相似性,将数据类别标签以层次化方式处理。但现有的类别间相似性度量均使用欧氏距离,由于欧氏距离无法有效处理高维数据,因此,受Tanimoto系数启发,提出一种新的类别相似性度量方法,使用Louvain算法构建树结构(TaniVT),考虑数据分布,设计基于类内散度的模糊粗糙分层分类器(fuzzy rough hierarchical classifier based on intra-class divergence,IDFRHC),将所提方法与已有的方法进行比较,通过实验验证了所提方法的有效性。展开更多
A new antenna selection algorithm for multiple input multiple output （MIMO） wireless systems is proposed. The modified Tanimoto coefficient is used to compare the similarity of the rows/columns of the channel matrix...A new antenna selection algorithm for multiple input multiple output （MIMO） wireless systems is proposed. The modified Tanimoto coefficient is used to compare the similarity of the rows/columns of the channel matrix. Based on the calculated similarity, the proposed algorithm chooses the antenna subset, which has the maximum product of dissimilarity and Frobenius norm. The proposed algorithm requires low computational complexity as to the optimal selection but with comparative outage capacity and average signal to noise ratio （SNR） performance. It can improve both the outage capacity and the average SNR as compared to random selection. The simulation results are shown to validate our algorithm.展开更多
Fidelity plays an important role in quantum information processing,which provides a basic scale for comparing two quantum states.At present,one of the most commonly used fidelities is Uhlmann-Jozsa(U-J)fidelity.Howeve...Fidelity plays an important role in quantum information processing,which provides a basic scale for comparing two quantum states.At present,one of the most commonly used fidelities is Uhlmann-Jozsa(U-J)fidelity.However,U-J fidelity needs to calculate the square root of the matrix,which is not trivial in the case of large or infinite density matrices.Moreover,U-J fidelity is a measure of overlap,which has limitations in some cases and cannot reflect the similarity between quantum states well.Therefore,a novel quantum fidelity measure called quantum Tanimoto coefficient(QTC)fidelity is proposed in this paper.Unlike other existing fidelities,QTC fidelity not only considers the overlap between quantum states,but also takes into account the separation between quantum states for the first time,which leads to a better performance of measure.Specifically,we discuss the properties of the proposed QTC fidelity.QTC fidelity is compared with some existing fidelities through specific examples,which reflects the effectiveness and advantages of QTC fidelity.In addition,based on the QTC fidelity,three discrimination coefficients d_(1)^(QTC),d_(2)^(QTC),and d_^(3)^(QTC)are defined to measure the difference between quantum states.It is proved that the discrimination coefficient d_(3)^(QTC)is a true metric.Finally,we apply the proposed QTC fidelity-based discrimination coefficients to measure the entanglement of quantum states to show their practicability.展开更多
In this paper we introduced Tanimoto based similarity measure for host-based intrusions using binary feature set for training and classification. The k-nearest neighbor (kNN) classifier has been utilized to classify a...In this paper we introduced Tanimoto based similarity measure for host-based intrusions using binary feature set for training and classification. The k-nearest neighbor (kNN) classifier has been utilized to classify a given process as either normal or attack. The experimentation is conducted on DARPA-1998 database for intrusion detection and compared with other existing techniques. The introduced similarity measure shows promising results by achieving less false positive rate at 100% detection rate.展开更多
针对逼近理想点排序法(technique for order preference by similarity to ideal solution,TOPSIS)存在的缺陷,提出基于Tanimoto系数和基于对称差的2种改进TOPSIS。改善或解决TOPSIS存在指标相关性问题、特殊样本集合无法比较优劣问题...针对逼近理想点排序法(technique for order preference by similarity to ideal solution,TOPSIS)存在的缺陷,提出基于Tanimoto系数和基于对称差的2种改进TOPSIS。改善或解决TOPSIS存在指标相关性问题、特殊样本集合无法比较优劣问题和样本数据动态变化时产生的逆序现象等缺陷;在稳定性、特异性、敏感性和有效性4方面对经典TOPSIS模型、改进Tanimoto模型和改进对称差模型进行对比验证,给出2种改进模型的适用场景。结果表明,2种方法各具有一定的优势。展开更多
动态时间规整(Dynamic Time Warping,DTW)是序列比对的经典方法,可以计算动态对应的序列间距的最小值.该文从一个新颖的角度重构了DTW的理论框架,提出了DTW的可加保优和始发保优两条特性.并依赖始发保优的算法拓展能力,将DTW由普通距离...动态时间规整(Dynamic Time Warping,DTW)是序列比对的经典方法,可以计算动态对应的序列间距的最小值.该文从一个新颖的角度重构了DTW的理论框架,提出了DTW的可加保优和始发保优两条特性.并依赖始发保优的算法拓展能力,将DTW由普通距离延拓到平均距离、Pearson相关系数、和Tanimoto相似系数的动态优化的计算模式,建立了一系列序列比对的新方法.动态计算的Pearson相关系数和Tanimoto相似系数较常规Pearson系数和Tanimoto系数分别更能捕捉长度一致的序列之间的真实相似性.对该系列方法以计算效率最高的动态平均距离为代表作内置,进行大量标注数据集的层次聚类比较测试,证实相较于传统DTW算法在基于序列比对的聚类准确率(平均F值)上至少有35个百分点的提高.且使用kNN作序列匹配分类实验也证实了动态平均距离优于传统DTW距离.展开更多
文摘在大数据时代,数据的类别标签数量激增,对现有的分类方法构成了重大挑战。为解决此问题,利用类别间的相似性,将数据类别标签以层次化方式处理。但现有的类别间相似性度量均使用欧氏距离,由于欧氏距离无法有效处理高维数据,因此,受Tanimoto系数启发,提出一种新的类别相似性度量方法,使用Louvain算法构建树结构(TaniVT),考虑数据分布,设计基于类内散度的模糊粗糙分层分类器(fuzzy rough hierarchical classifier based on intra-class divergence,IDFRHC),将所提方法与已有的方法进行比较,通过实验验证了所提方法的有效性。
文摘A new antenna selection algorithm for multiple input multiple output （MIMO） wireless systems is proposed. The modified Tanimoto coefficient is used to compare the similarity of the rows/columns of the channel matrix. Based on the calculated similarity, the proposed algorithm chooses the antenna subset, which has the maximum product of dissimilarity and Frobenius norm. The proposed algorithm requires low computational complexity as to the optimal selection but with comparative outage capacity and average signal to noise ratio （SNR） performance. It can improve both the outage capacity and the average SNR as compared to random selection. The simulation results are shown to validate our algorithm.
基金supported by the National Natural Science Foundation of China(62003280,61976120)Chongqing Talents:Exceptional Young Talents Project(cstc2022ycjh-bgzxm0070)+2 种基金Natural Science Foundation of Chongqing(2022NSCQ-MSX2993)Natural Science Key Foundation of Jiangsu Education Department(21KJA510004)Chongqing Overseas Scholars Innovation Program(cx2022024)。
文摘Fidelity plays an important role in quantum information processing,which provides a basic scale for comparing two quantum states.At present,one of the most commonly used fidelities is Uhlmann-Jozsa(U-J)fidelity.However,U-J fidelity needs to calculate the square root of the matrix,which is not trivial in the case of large or infinite density matrices.Moreover,U-J fidelity is a measure of overlap,which has limitations in some cases and cannot reflect the similarity between quantum states well.Therefore,a novel quantum fidelity measure called quantum Tanimoto coefficient(QTC)fidelity is proposed in this paper.Unlike other existing fidelities,QTC fidelity not only considers the overlap between quantum states,but also takes into account the separation between quantum states for the first time,which leads to a better performance of measure.Specifically,we discuss the properties of the proposed QTC fidelity.QTC fidelity is compared with some existing fidelities through specific examples,which reflects the effectiveness and advantages of QTC fidelity.In addition,based on the QTC fidelity,three discrimination coefficients d_(1)^(QTC),d_(2)^(QTC),and d_^(3)^(QTC)are defined to measure the difference between quantum states.It is proved that the discrimination coefficient d_(3)^(QTC)is a true metric.Finally,we apply the proposed QTC fidelity-based discrimination coefficients to measure the entanglement of quantum states to show their practicability.
文摘In this paper we introduced Tanimoto based similarity measure for host-based intrusions using binary feature set for training and classification. The k-nearest neighbor (kNN) classifier has been utilized to classify a given process as either normal or attack. The experimentation is conducted on DARPA-1998 database for intrusion detection and compared with other existing techniques. The introduced similarity measure shows promising results by achieving less false positive rate at 100% detection rate.
文摘针对逼近理想点排序法(technique for order preference by similarity to ideal solution,TOPSIS)存在的缺陷,提出基于Tanimoto系数和基于对称差的2种改进TOPSIS。改善或解决TOPSIS存在指标相关性问题、特殊样本集合无法比较优劣问题和样本数据动态变化时产生的逆序现象等缺陷;在稳定性、特异性、敏感性和有效性4方面对经典TOPSIS模型、改进Tanimoto模型和改进对称差模型进行对比验证,给出2种改进模型的适用场景。结果表明,2种方法各具有一定的优势。
文摘动态时间规整(Dynamic Time Warping,DTW)是序列比对的经典方法,可以计算动态对应的序列间距的最小值.该文从一个新颖的角度重构了DTW的理论框架,提出了DTW的可加保优和始发保优两条特性.并依赖始发保优的算法拓展能力,将DTW由普通距离延拓到平均距离、Pearson相关系数、和Tanimoto相似系数的动态优化的计算模式,建立了一系列序列比对的新方法.动态计算的Pearson相关系数和Tanimoto相似系数较常规Pearson系数和Tanimoto系数分别更能捕捉长度一致的序列之间的真实相似性.对该系列方法以计算效率最高的动态平均距离为代表作内置,进行大量标注数据集的层次聚类比较测试,证实相较于传统DTW算法在基于序列比对的聚类准确率(平均F值)上至少有35个百分点的提高.且使用kNN作序列匹配分类实验也证实了动态平均距离优于传统DTW距离.
