基于随机投影与集成学习的离群点检测算法

针对传统基于相似度的离群点检测算法在高维不均衡数据集上效果不够理想的问题,提出一种新颖的基于随机投影与集成学习的离群点检测（ensemble learning and random projection-based outlier detection,EROD）框架。算法首先集成多个随机投影方法对高维数据进行降维,提升数据多样性;然后集成多个不同的传统离群点检测器构建异质集成模型,增加算法鲁棒性;最后使用异质模型对降维后的数据进行训练,训练后的模型经过两次优化组合以降低泛化误差,输出最终的对象离群值,离群值高的对象被算法判定为离群点。分别在四个不同领域的高维不均衡真实数据集上进行对比实验,结果表明该算法与传统离群点检测算法和基于集成学习的离群点检测算法相比,在AUC和precision@n值上平均提高了3.6%和14.45%,证明EROD算法具有处理高维不均衡数据异常的优势。     

基于自编码器与集成学习的离群点检测算法

针对基于自编码器的离群点检测算法在中小规模数据集上易过拟合以及传统的基于集成学习的离群点检测算法未对基检测器进行优化选择而导致的检测精度低的问题,提出了一种基于自编码器与集成学习的离群点检测（EAOD）算法。首先,随机改变自编码器的连接结构来生成不同的基检测器,以获取数据对象的离群值和标签离群值;然后,通过最近邻算法计算数据对象之间的欧氏距离,并在对象周围构建局部区域;最后,根据离群值与标签离群值之间的相似度,选择在该区域内检测能力强的基检测器进行组合,组合后的对象离群值作为EAOD算法最终判定的离群值。在实验中,所提算法与自编码器（AE）算法相比,在Cardio数据集上,接受者操作特征曲线下方的面积（AUC）和平均精度（AP）分值分别提高了8.08个百分点和9.17个百分点;所提算法与特征装袋（FB）集成学习算法相比,在Mnist数据集上,运行时间成本降低了21.33%。实验结果表明,在无监督学习下所提算法具有良好的检测性能和检测实时性。     

基于图上随机游走的离群点检测算法

离群点检测算法在网络入侵检测、医疗辅助诊断等领域具有十分广泛的应用。针对LDOF、CBOF及LOF算法在大规模数据集和高维数据集的检测过程中存在的执行时间长及检测率较低的问题,提出了基于图上随机游走(BGRW)的离群点检测算法。首先初始化迭代次数、阻尼因子以及数据集中每个对象的离群值;其次根据对象之间的欧氏距离推导出漫步者在各对象之间的转移概率;然后通过迭代计算得到数据集中每个对象的离群值;最后将数据集中离群值最高的对象判定为离群点并输出。在UCI真实数据集与复杂分布的合成数据集上进行实验,将BGRW算法与LDOF、CBOF和LOF算法在执行时间、检测率和误报率指标上进行对比。实验结果表明,BGRW算法能够有效降低执行时间并在检测率及误报率指标上优于对比算法。     

基于空间局部偏离因子的离群点检测算法

针对空间数据集的特性,提出一种基于空间局部偏离因子(SLDF)的离群点检测算法。利用SLDF度量空间点对象的离群程度,计算空间数据集中点对象的SLDF值并对其进行排序,将取值较大的前M个点对象作为空间离群点。实验结果表明,该算法能较好地检测空间局部离群点,其有效性与准确性均优于SLZ算法,适用于高维大数据集的空间离群点检测。     

基于KNN图的空间离群点挖掘算法

空间数据集中离群数据与正常数据之间的非空间属性值相差较大。针对该情况,提出一种基于K-最邻近(KNN)图的空间离群点挖掘算法。该算法通过所有对象的K近邻关系构造KNN图,将相邻对象非空间属性值的差作为2个对象点间的边权值,利用裁边策略去掉权值较高的边,从而识别出空间离群点和离群区域。实验结果表明,该算法的时间性能优于POD算法。     

生成对抗网络GAN综述

生成对抗网络(GAN)作为一种新的无监督学习算法框架得到越来越多研究者的青睐,已然成为当下的一个研究热点。GAN受启发于博弈论中的二人零和博弈理论,其独特的对抗训练思想能生成高质量的样本,具有比传统机器学习算法更加强大的特征学习和特征表达能力。目前GAN在计算机视觉领域尤其是在样本生成领域取得显著成功,每年有大量GAN相关研究的论文产出。针对GAN这一热点模型,首先介绍了GAN的研究现状;接着介绍了GAN的理论、框架,详细分析了GAN在训练过程中存在梯度消失和模式崩溃的原因;然后讨论了一些典型的GAN的改进模型,总结了它们理论的改进之处、优点、局限性、应用场景以及实现成本,同时还将GAN与VAE、RBM模型进行比较,总结出GAN的优势和劣势;最后展示了GAN在数据生成、图像超分辨率、图像风格转换等方面的应用成果,并探讨了GAN目前面临的挑战以及未来的研究方向。     

基于密度划分的离群点检测算法

目前,大部分离群点检测算法需要人工输入参数,不能同时检测出全局和局部离群点,不能有效处理密度不均匀数据。针对这些问题,提出一种基于密度划分的离群点检测算法DD-DBSCAN。主要创新包括:1)运用最小生成树的方法,新定义簇密度概念,将数据录入后划分成密度不等的簇,使算法能够处理密度分布不均匀的数据;2)采用"分而治之"的思想,对经过划分的数据集分别进行离群点检测,使得算法能够同时处理全局和局部离群点;3)通过在各个簇中自适应地计算所需参数值,算法不再需要人工输入参数(聚类半径(Eps)等)。通过在2D模拟数据集和Iris真实数据集上的实验表明,与DBSCAN算法比较,本文算法具有更高的覆盖率和正确率。     

基于局部信息熵的加权子空间离群点检测算法

离群点检测作为数据挖掘的一个重要研究方向,可以从大量数据中发现少量与多数数据有明显区别的数据对象,"维度灾殃"现象的存在使得很多已有的离群点检测算法对高维数据不再有效,针对这一问题,提出基于局部信息熵的加权子空间离群点检测算法SPOD,通过对数据对象在各维进行邻域信息熵分析,生成数据对象相应的离群子空间和属性权向量,对离群子空间中的属性赋以较高的权值,进一步提出子空间加权距离等概念,采用基于密度离群点检测的思想,分析计算数据对象的子空间离群影响因子,判断是否为离群点,算法能够有效地适应于高维数据离群点检测,理论分析和实验结果表明算法是有效可行的.     

基于近邻传播的离群点检测算法

离群点是与其他正常点属性不同的一类对象,其检测技术在各行业上均有维护数据纯度、保障业内安全等重要应用,现有算法大多是基于距离、密度等传统方法判断检测离群点.本算法给每个对象分配一个"孤立度",即该点相对其邻点的孤立程度,通过排序进行判定,比传统算法效率更高.在AP(affinity propagation)聚类算法的基础上进行改进与优化,提出能检测异常数据点的算法APO(outlier detection algorithm based on affinity propagation).通过加入孤立度模块并计算处理样本点的孤立信息,并引入放大因子,使其与正常点之间的差异更明显,通过增大算法对离群点的敏感性,提高算法的准确性.分别在模拟数据集和真实数据集上进行对比实验,结果表明:该算法与AP算法相比,对离群点的敏感性更加强烈,且本算法检测离群点的同时也能聚类,是其他检测算法所不具备的.     

基于多示例学习的局部离群点改进算法

在多示例学习框架下,训练数据集由若干个包组成,包内含有多个用属性-值对形式表示的示例,系统对包内的多个示例进行学习。传统的基于多示例学习的局部离群点检测算法将多示例学习框架运用到数据集上,将多示例问题转化为单示例问题进行处理。但在示例包的转换过程中采用示例内部的特征长度所占比作为权重机制,并没有考察对结果影响较大的示例,分析原因或者动态调整其权重,从而对离群点检测的效果造成影响。针对这一问题,为了充分适应数据内部的分布特征,提出了一种基于多示例学习的局部离群点改进算法FWMIL-LOF。算法采用MIL（Multi-Instance Learning）框架,在示例包的转换过程中引入描述数据重要度的权重函数,通过定义惩罚策略对权重函数做相应调整,从而确定了不同特征属性的示例在所属包中的权重。在实际企业的实时采集监控系统中,通过仿真分析,并与其他经典局部离群点检测算法进行对比,验证了改进算法在离群点检测效果方面的提高。     

Resolution-based outlier factor: detecting the top-n most outlying data points in engineering data

One of the common endeavours in engineering applications is outlier detection, which aims to identify inconsistent records from large amounts of data. Although outlier detection schemes in data mining discipline are acknowledged as a more viable solution to efficient identification of anomalies from these data repository, current outlier mining algorithms require the input of domain parameters. These parameters are often unknown, difficult to determine and vary across different datasets containing different cluster features. This paper presents a novel resolution-based outlier notion and a nonparametric outlier-mining algorithm, which can efficiently identify and rank top listed outliers from a wide variety of datasets. The algorithm generates reasonable outlier results by taking both local and global features of a dataset into account. Experiments are conducted using both synthetic datasets and a real life construction equipment dataset from a large road building contractor. Comparison with the current outlier mining algorithms indicates that the proposed algorithm is more effective and can be integrated into a decision support system to serve as a universal detector of potentially inconsistent records.     

Outlier detection by example

Outlier detection is a useful technique in such areas as fraud detection, financial analysis and health monitoring. Many recent approaches detect outliers according to reasonable, pre-defined concepts of an outlier (e.g., distance-based, density-based, etc.). However, the definition of an outlier differs between users or even datasets. This paper presents a solution to this problem by including input from the users. Our OBE (Outlier By Example) system is the first that allows users to provide examples of outliers in low-dimensional datasets. By incorporating a small number of such examples, OBE can successfully develop an algorithm by which to identify further outliers based on their outlierness. Several algorithmic challenges and engineering decisions must be addressed in building such a system. We describe the key design decisions and algorithms in this paper. In order to interact with users having different degrees of domain knowledge, we develop two detection schemes: OBE-Fraction and OBE-RF. Our experiments on both real and synthetic datasets demonstrate that OBE can discover values that a user would consider outliers.     

Fast mining of distance-based outliers in high-dimensional datasets

Defining outliers by their distance to neighboring data points has been shown to be an effective non-parametric approach to outlier detection. In recent years, many research efforts have looked at developing fast distance-based outlier detection algorithms. Several of the existing distance-based outlier detection algorithms report log-linear time performance as a function of the number of data points on many real low-dimensional datasets. However, these algorithms are unable to deliver the same level of performance on high-dimensional datasets, since their scaling behavior is exponential in the number of dimensions. In this paper, we present RBRP, a fast algorithm for mining distance-based outliers, particularly targeted at high-dimensional datasets. RBRP scales log-linearly as a function of the number of data points and linearly as a function of the number of dimensions. Our empirical evaluation demonstrates that we outperform the state-of-the-art algorithm, often by an order of magnitude.     

加速大规模数据集的离群点检测

针对现有基于距离的离群点检测算法在处理大规模数据时效率低的问题,提出一种基于聚类和索引的分布式离群点检测(DODCI) 算法。首先利用聚类方法将大数据集划分成簇;然后在分布式环境中的各节点处并行创建各个簇的索引;最后使用两个优化策略和两条剪枝规则以循环的方式在各节点处进行离群点检测。在合成数据集和整理后的KDD CUP数据集上的实验结果显示,在数据量较大时该算法比Orca和iDOoR算法快近一个数量级。理论和实验分析表明,该算法可以有效提高大规模数据中离群点的检测效率。     

Outlier detection in skewed data

In this work we develop two new algorithms for outlier detection in skewed data. The first algorithm uses an adjusted median with the help of Robust Support Vector Regression and the second one estimates a robust covariance matrix and incorporates it in the Kernel Density Estimation of the data. We applied these two methods in outlier detection in univariate skewed and multivariate skewed data and show that they succed in detecting the outliers and also in including more inliers. Based on simulation study it is realized that the proposed procedures have the advantage in skewed data. Finally, both methods are also used in detecting MRIs of patients diagnosed with dementia by measuring the outlyingness from the MRIs of nondemented patients.     

基于自编码标准流的异常点检测

在大型和高维数据上进行有效检测, 在实际应用中具有重要意义. 异常点检测是指识别出偏离一般数据分布的数据点, 其核心是密度估计. 尽管像深度自编码高斯混合模型通过先降低维度, 再进行密度估计已经取得了重大进展, 但是它对低维潜在空间引入噪声, 并且在对密度估计模块优化时存在一些限制, 例如需要保证协方差是正定矩阵. 为解决这些限制, 本文提出一种用于无监督异常检测的深度自编码标准化流(deep autoencoder normalizing flow, DANF). 该模型利用深度自编码器为每个输入样本生成低维潜在空间表示和重构误差, 进而将其输入标准化流(normalizing flow, NF), 最终映射成高斯分布. 在多个公开的基准数据集上的实验结果表明, 深度自编码标准化流模型显著优于最先进的异常检测技术, 在评估指标F1-score上最高提升26.43%.     

Distance-based detection and prediction of outliers

A distance-based outlier detection method that finds the top outliers in an unlabeled data set and provides a subset of it, called outlier detection solving set, that can be used to predict the outlierness of new unseen objects, is proposed. The solving set includes a sufficient number of points that permits the detection of the top outliers by considering only a subset of all the pairwise distances from the data set. The properties of the solving set are investigated, and algorithms for computing it, with subquadratic time requirements, are proposed. Experiments on synthetic and real data sets to evaluate the effectiveness of the approach are presented. A scaling analysis of the solving set size is performed, and the false positive rate, that is, the fraction of new objects misclassified as outliers using the solving set instead of the overall data set, is shown to be negligible. Finally, to investigate the accuracy in separating outliers from inliers, ROC analysis of the method is accomplished. Results obtained show that using the solving set instead of the data set guarantees a comparable quality of the prediction, but at a lower computational cost.     

GridOF:面向大规模数据集的高效离群点检测算法

作为数据库知识发现研究的重要技术手段，现有离群点检测算法在运用于大型数据集时其时间与空间效率均无法令人满意．通过对数据集中离群点分布特征的分析，在数据空间网格划分的基础上，研究数据超方格层次上的密度近似计算与稠密数据主体滤除策略．给出通过简单的修正近似计算取代繁复的点对点密度函数值计算的方法．基于上述思想构造的离群点检测算法GlidOF在保持足够检测精度的同时显著降低了时空复杂度，运用于大规模数据集离群点检测具有良好的适用性和有效性．     

ODIC-DBSCAN:一种新的簇内孤立点分析算法

长期以来,孤立点的检测一直聚焦于簇边缘的离散点,当聚类后簇的数目低于实际数目,或孤立点被伪装在簇内的情况下,簇内孤立点的判定则会更加困难.为判定簇内孤立点,提出一种基于密度聚类DBSCAN（Density based spatial clustering of application with noise）的簇内孤立点检测方法ODIC-DBSCAN（Outlier detection of inner-cluster based on DBSCAN）.首先在建立距离矩阵的基础上,通过半径获取策略得到针对该点集的k个有效半径Radius集合,并据此构造密度矩阵;然后建立点集覆盖模型,提出了相邻有效半径构造的覆盖多维体能够覆盖点集的思想,并通过拉格朗日乘子法求取最优的覆盖多维体数目之比,输出点比阈值组;最后重建ODIC-DBSCAN的孤立点检测方法,以簇发生融合现象作为算法终止的判定条件.实验通过模拟数据集,公开benchmark与UCI数据集共同验证了ODIC-DBSCAN算法,展示了聚类过程;分析了算法性能;并通过与其他聚类、孤立点判定方法的对比,验证了算法对簇内孤立点的判定效果.