摔倒检测中的样本失衡问题研究

由于真实的摔倒数据难以获得,导致采集到的正常行为和摔倒行为样本比例严重失衡,从而基于此数据集训练的常规摔倒检测模型的漏警率和误警率都较高,不能满足实际的需求。针对该问题,提出一种基于样本加权极速学习机的摔倒检测方法,该方法综合考虑不同种类行为样本之间的比例关系,分别赋予其一定的权值,能较好地解决样本失衡问题。基于真实行为数据的实验结果表明,和传统非加权的行为识别方法相比较,基于样本加权极速学习机的摔倒检测方法能够将识别模型的性能提高10%左右。     

基于机器学习算法的主机恶意代码检测技术研究

对机器学习算法下主机恶意代码检测的主流技术途径进行了研究,分别针对静态、动态这2种分析模式下的检测方案进行了讨论,涵盖了恶意代码样本采集、特征提取与选择、机器学习算法分类模型的建立等要点。对机器学习算法下恶意代码检测的未来工作与挑战进行了梳理。为下一代恶意代码检测技术的设计和优化提供了重要的参考。     

Guaranteed detection of an imbalance instant of the GARCH-process

A detection instant problem of the GARCH-process parametric change with random noise distribution is studied. A sequential procedure of imbalance instant detection, where a decision of change existence is made at a given instant of time after the accumulation of given distinctions between possible process models, is suggested. Formulas for procedure characteristics calculation are developed, asymptotic qualities of the procedure are studied. Model estimated values are presented.     

Interest point detection using imbalance oriented selection

Interest point detection has a wide range of applications, such as image retrieval and object recognition. Given an image, many previous interest point detectors first assign interest strength to each image point using a certain filtering technique, and then apply non-maximum suppression scheme to select a set of interest point candidates. However, we observe that non-maximum suppression tends to over-suppress good candidates for a weakly textured image such as a face image. We propose a new candidate selection scheme that chooses image points whose zero-/first-order intensities can be clustered into two imbalanced classes (in size), as candidates. Our tests of repeatability across image rotations and lighting conditions show the advantage of imbalance oriented selection. We further present a new face recognition application—facial identity representability evaluation—to show the value of imbalance oriented selection.     

Training cost-sensitive neural networks with methods addressing the class imbalance problem

This paper studies empirically the effect of sampling and threshold-moving in training cost-sensitive neural networks. Both oversampling and undersampling are considered. These techniques modify the distribution of the training data such that the costs of the examples are conveyed explicitly by the appearances of the examples. Threshold-moving tries to move the output threshold toward inexpensive classes such that examples with higher costs become harder to be misclassified. Moreover, hard-ensemble and soft-ensemble, i.e., the combination of above techniques via hard or soft voting schemes, are also tested. Twenty-one UCl data sets with three types of cost matrices and a real-world cost-sensitive data set are used in the empirical study. The results suggest that cost-sensitive learning with multiclass tasks is more difficult than with two-class tasks, and a higher degree of class imbalance may increase the difficulty. It also reveals that almost all the techniques are effective on two-class tasks, while most are ineffective and even may cause negative effect on multiclass tasks. Overall, threshold-moving and soft-ensemble are relatively good choices in training cost-sensitive neural networks. The empirical study also suggests that some methods that have been believed to be effective in addressing the class imbalance problem may, in fact, only be effective on learning with imbalanced two-class data sets.     

基于Win32 API的未知病毒检测

提出了一个基于行为特征向量的病毒检测方法。特征向量的每一维用于表示一种恶意行为事件,每一事件由相应的Win32应用程序编程接口(API)调用及其参数表示,并实现了一个自动化行为追踪系统(Argus)用于行为特征的提取。试验中,通过对样本数据的分析,利用互信息对特征向量进行属性约简,减少特征维数。试验结果表明,约简后的模型对于发生行为事件数大于1的病毒程序仍有着较好的检测效果。     

Decision tree induction based on minority entropy for the class imbalance problem

Most well-known classifiers can predict a balanced data set efficiently, but they misclassify an imbalanced data set. To overcome this problem, this research proposes a new impurity measure called minority entropy, which uses information from the minority class. It applies a local range of minority class instances on a selected numeric attribute with Shannon’s entropy. This range defines a subset of instances concentrating on the minority class to be constructed by decision tree induction. A decision tree algorithm using minority entropy shows improvement compared with the geometric mean and F-measure over C4.5, the distinct class-based splitting measure, asymmetric entropy, a top–down decision tree and Hellinger distance decision tree on 24 imbalanced data sets from the UCI repository.     

Solving the collision detection problem

Considers how a happy convergence of factors in the mid 195Os led CEIT and the University of Navarre to team up for an ambitious project. First, we accumulated expertise in the fields of computer graphics, mechanism analysis and solid modeling. Second, the appearance of 3D graphics workstations on the market allowed us to put our experience to work to develop CompAMM (Computer Analysis of Machines and Mechanisms), a general-purpose program to simulate and visualize in real time the kinematic analysis of complex articulated bodies. We got a Hewlett Packard 350 SRX in July 1987 and made our first presentation of the real-time analysis and visualization of 3D mechanisms at an international congress in Seville two months later. Target applications included robotics, vehicles, mechanisms, spaceship manoeuvres, teleoperator training, astronaut motion, and ergonomics inside vehicles     

基于模糊模式识别的未知病毒检测

提出了一种基于模糊模式识别的检测方法来实现对计算机病毒的近似判别。该方法可以克服病毒特征代码扫描法不能识别未知病毒的缺点。在该检测方法的基础上,文中设计了一个病毒检测网络模型,此模型既可以实现对已知病毒的查杀,又可以对可疑程序行为进行分析评判,最终实现对未知病毒的识别。     

基于Win32 API和SVM的未知病毒检测方法

提出了一种Windows平台下检测未知病毒的新方法,该方法通过分析PE文件调用的Win32 API序列,用SVM来对划分后k长度的API短序列分类,并通过分析API函数及参数危险程度来提高SVM分类的精确度,从而实现对未知病毒的检测。实验结果表明,该方法实现的病毒检测系统比只用SVM的系统具有更好的检测效果。     

基于API序列分析和支持向量机的未知病毒检测

提出了一种在Windows平台下检测未知病毒的新方法,以PE文件调用的WinAPI序列为特征,运用支持向量机分类来检测未知病毒。实验结果表明,所实现BK 50系统对未知病毒具有较好的识别效果。     

不平衡数据集下特征词两面性的新型降维算法

传统DFS特征选择算法在降维处理时既未考虑样本分布不均的情况,又未涉及负特征词对类别的影响。本文综合考虑DFS的缺陷并进行优化处理,将DFS与卡方检测算法CHI结合,提出一种改进型特征选择算法DFS-sCHI,引入负特证词作为类别划分的影响因子之一,解决不平衡数据集下所提特征词类别分布不均的问题。经实验分析,不平衡数据集下,DFS-sCHI相比较于DFS,在分类精度上有明显提高。     

On the disturbed fault detection and isolation problem

In this paper we study the fault detection and isolation problem in presence of disturbances. In the case of observer-based residual generation, the problem amounts to finding two gain matrices such that two problems are simultaneously solved. These problems are insensitivity of the residuals to disturbances and the existence of some special structure for the transfer from faults to residuals. We prove in this paper that this joint problem can be solved if and only if the usual (undisturbed) fault detection and isolation problem can be solved for a system with a reduced number of outputs.     

Equation of SVM-rebalancing: the point-normal form of a plane for class imbalance problem

The task of detecting a rare but important class has extensively been studied in the machine learning community. It is commonly agreed that traditional classifiers are certainly limited to imbalanced datasets and do not perform well. A number of solutions to the problem were proposed at both data and algorithmic levels. We propose the point-normal form of a plane, namely SVM-rebalancing, to be based on the second type. In this learning process, the assumption of pseudo-prior probabilities provides a rebalanced recipe for countering the imbalance inspired by Bayesian decision theory. Thus, we set a rebalancing programming problem by incorporating a rebalanced heuristics into the fitting of model to raise the class separability. In addition, various measures are used to characterize the performance of classifiers. Compared with several popular decision tree splitting criteria and cost-sensitive learning, the proposed method gives comparable separability with minority class to avoid heavy biasing of the majority class.

     

Handling class imbalance problem in software maintainability prediction: an empirical investigation

As the complexity of software systems is increasing; software maintenance is becoming a challenge for software practitioners. The prediction of classes that require high maintainability effort is of utmost necessity to develop cost-effective and high-quality software. In research of software engineering predictive modeling, various software maintainability prediction (SMP) models are evolved to forecast maintainability. To develop a maintainability prediction model, software practitioners may come across situations in which classes or modules requiring high maintainability effort are far less than those requiring low maintainability effort. This condition gives rise to a class imbalance problem (CIP). In this situation, the minority classes’ prediction, i.e., the classes demanding high maintainability effort, is a challenge. Therefore, in this direction, this study investigates three techniques for handling the CIP on ten open-source software to predict software maintainability. This empirical investigation supports the use of resampling with replacement technique (RR) for treating CIP and develop useful models for SMP.     

Inverse random under sampling for class imbalance problem and its application to multi-label classification

In this paper, a novel inverse random under sampling (IRUS) method is proposed for the class imbalance problem. The main idea is to severely under sample the majority class thus creating a large number of distinct training sets. For each training set we then find a decision boundary which separates the minority class from the majority class. By combining the multiple designs through fusion, we construct a composite boundary between the majority class and the minority class. The proposed methodology is applied on 22 UCI data sets and experimental results indicate a significant increase in performance when compared with many existing class-imbalance learning methods. We also present promising results for multi-label classification, a challenging research problem in many modern applications such as music, text and image categorization.     

Lower bounds for the cycle detection problem

Given a function f over a domain and an element x in the domain, the cycle detection problem is to find a repetition in the sequence of values x, f(x), f(f(x)), f³(x),…, if one exists. This paper investigates lower bounds on the number of function evaluations needed when there is a bound on the amount of memory available. For certain restricted classes of algorithms which use two memory locations optimality is achieved. A summary of the major results appears in the final section.     

改进的SVM解决背景知识数据中的类不平衡*

针对背景知识数据集中存在的类不平衡对分类器的影响,根据背景知识数据集样本量小、数据维数高的特性分析了目前各种方法在解决背景知识数据中的类不平衡问题时的缺陷,提出了一种基于分类后处理的改进SVM算法。改进算法引入权重参数调整SVM的分类决策函数,提高少类样本对分类器的贡献,使分类平面向多类样本倾斜,从而解决类不平衡对SVM造成的影响。在MAROB数据集上的实验表明,改进算法对少类的预测效果要优于传统的机器学习算法。     

The decentralized quickest detection problem

Two detectors making independent observations must decide when a Markov chain jumps from state 0 to state 1. The decisions are coupled through a common cost function. It is shown that the optimal decision is characterized by thresholds as in the decoupled case. However, the thresholds are time-varying and their computation requires the solution of two coupled sets of dynamic programming equations. A comparison to the decoupled case shows the structure of the coupling.