融合改进二元萤火虫算法和互补性测度的集成剪枝方法

差异性和平均精度是提高分类器集成性能的两个重要指标。增加差异性势必会降低平均精度,增大平均精度一定会减小差异性。故在差异性和平均精度之间存在一个平衡状态,使得集成性能最优。为了寻找该平衡状态,该文提出融合改进二元萤火虫算法和互补性测度的集成剪枝方法。首先,采用bootstrap抽样方法独立训练出多个基分类器,构建原始基分类器池。其次,采用互补性测度对原始基分类器池进行预剪枝。接着,通过改进萤火虫的移动方式和搜索过程,引入重新初始化机制和跳跃行为,提出改进二元萤火虫算法。最后,采用改进二元萤火虫算法对预剪枝后的基分类器,进行进一步剪枝,选择出集成性能最优的基分类器子集合。在5个UCI数据集上的实验结果表明,较其他方法,使用较少的基分类器,获得了更优的集成性能,具有良好的有效性和显著性。     

基于半监督协同训练的无人机对地目标跟踪

由于目标小、可区分性差,无人机对地目标跟踪较传统视频目标跟踪更容易丢失目标,提出一种基于l1图半监督协同训练的目标跟踪算法。算法首先提取样本的颜色和纹理特征构建两个充分冗余的视图,再以基于l1图的半监督学习算法取代传统协同训练中的监督学习方法构建单视图中的分类器,提高有限标记样本条件下的分类正确率,然后通过基于负类学习的协同训练算法协同更新两个视图的分类器,最后根据不同视图的相似度分布熵融合各分类器的分类结果实现目标跟踪。实验结果表明,该算法能够有效提高分类器的判别能力,具有良好的跟踪性能。     

一种基于类Haar特征和改进AdaBoost分类器的车辆识别算法

 提出一种基于类haar特征和改进AdaBoost分类器的车辆图像识别算法,以解决当前基于SVM分类器或级联分类器存在的分类识别性能不足以及传统基于AdaBoost算法的训练所需时间过长的问题.首先,基于积分图提取图像的扩展类haar特征,然后对所提取的海量类haar特征应用改进的AdaBoost分类器训练方法进行特征选择及分类器训练,最后利用所选择的特征信息及训练得到的分类器进行两类分类识别.实验结果表明,文中方法无论是在识别性能还是训练所需时间方面均明显优于传统方法,具有较好的应用前景.     

基于图的Co-Training网页分类

 本文充分利用网页数据的超链接关系和文本信息,提出了一种用于网页分类的归纳式半监督学习算法:基于图的Co-training网页分类算法(Graph based Co-training algorithm for web page classification),简称GCo-training,并从理论上证明了算法的有效性.GCo-training在Co-training算法框架下,迭代地学习一个基于由超链接信息构造的图的半监督分类器和一个基于文本特征的Bayes 分类器.基于图的半监督分类器只利用少量的标记数据,通过挖掘数据间大量的关系信息就可达到比较高的预测精度,可为Bayes分类器提供大量的标记信息;反过来学习大量标记信息后的Bayes分类器也可为基于图的分类器提供有效信息.迭代过程中,二者互相帮助,不断提高各自的性能,而后Bayes分类器可以用来预测大量未见数据的类别.在Web→KB数据集上的实验结果表明,与利用文本特征和锚文本特征的Co-training算法和基于EM的Bayes算法相比,GCo-training算法性能优越.     

基于随机子空间和AdaBoost的自适应集成方法

如何构造差异性大且精确度高的基分类器是集成学习的重点,为此提出一种新的集成学习方法——利用PSO寻找使得AdaBoost依样本权重抽取的数据集分类错误率最小化的最优特征权重分布,依据此最优权重分布对特征随机抽样生成随机子空间,并应用于AdaBoost的训练过程中.这就在增加分类器间差异性的同时保证了基分类器的准确度.最后用多数投票法融合各基分类器的决策结果,并通过仿真实验验证该方法的有效性.     

基于KPCA和RBF网络的文本分类研究

基于词空间的分类方法很难处理文本的高维特性和复杂相关性,为此文中提出了基于核的主成分分析和径向基神经网络的文本分类算法.首先利用核主成分分析选择合适的核函数从高维特征空间中提取文本向量的主成分,实现了文本输入空间的降维和语义特征空间的抽取,然后在语义特征空间中训练径向基神经网络分类器,并利用训练得到的分类器进行文本分类工作.实验结果表明:核主成分分析不仅实现了降维,而且能在大幅减减少径向基神经网络训练时间的基础上显著提高其分类精度.     

基于Boosting算法的车牌汉字识别

对AdaBoost算法作了详细的分析:AdaBoost是一种有效的分类器组合方法,他用某个分类算法生成一系列的基分类器,每个基分类器的训练依赖于在其之前产生的分类器的分类结果,基分类器在训练集上的错误率用于调整训练样本的概率分布,最终分类器通过单个基分类器的加权投票建立起来。最后将该算法应用于车牌照的汉字识别,对识别效率有大幅度提高。     

一种增强差异性的半监督协同分类算法

半监督学习中的Tri-Training算法打破了以往算法对充分冗余视图的限制,并通过利用三个分类器处理标记置信度和样本预测问题提高了标记效率.为进一步增强协同训练过程中分类器之间的差异性以提高性能,本文在其理论基础上提出了一种增强差异性的半监督协同分类算法.该算法利用三个不同的分类器进行学习;考虑到分类模型在更新过程中,可能会因随机抽样导致性能恶化,该算法利用基于标记类别的分层抽样法来对已标记样本集进行抽样,并通过基于分类正确率的加权投票法实现了分类器的集成,提高了预测准确率.本文通过实验对所提出算法与Tri-Training算法做了性能比较,实验结果表明本文所提出的方法在分类问题上具有较好的性能,验证了该算法的有效性和可行性.     

基于信息熵差异性度量的数据流增量集成分类算法

对分类器之间的差异性进行了研究,提出了一种基于信息熵差异性度量的增量集成分类算法,将信息熵差异性度量方法融入到基分类器选择过程中,通过对训练数据集的基分类结果的信息熵差异度计算,采用循环迭代优化的选择方法,以熵差异性最优化为约束目标,动态调整基分类器个数,实现了分类准确稳定,减少了系统开销。通过实验比对,证明了算法在数据流处理时比其他算法具有更小的开销和较强的适应性。     

基于K-medoids聚类的贝叶斯集成算法

朴素贝叶斯分类算法由于其计算高效在生活中应用广泛。本文根据集成算法的差异性特征,聚类算法聚类点的选择方式的可变性,提出了基于K-medoids聚类技术的贝叶斯集成算法,朴素贝叶斯的泛化性能得到了提升。首先,通过样本集训练出多个朴素贝叶斯基分类器模型;然后,为了增大基分类器之间的差异性,利用K-medoids算法对基分类器在验证集上的预测结果进行聚类;最后,从每个聚类簇中选择泛化性能最佳的基分类器进行集成学习,最终结果由简单投票法得出。将该算法应用于UCI数据集,并与其他类似算法进行比较可得,本文提出的基于K-medoids聚类的贝叶斯集成算法(NBKME)提高了数据集的分类准确率。     

Ground bounce modelling for digital gigascale integrated circuits

In the paper, an analytical model for ground bounce noise evaluation taking into account the interdependence between I_DD switching current and V_DD noise voltage is presented. The model shows the discrepancies from general accepted assumption of independence between the two variables. The main conclusion is that noise calculations using the independence assumption cause an overestimation of the noise levels. The results are verified through realistic simulations and for different technology nodes and accurate analysis of two canonical circuits.     

The good, bad and ugly: distributed detection of a known signal in dependent Gaussian noise

Most results about quantized detection rely strongly on an assumption of independence among random variables. With this assumption removed, little is known. Thus, in this paper, Bayes-optimal binary quantization for the detection of a shift in mean in a pair of dependent Gaussian random variables is studied. This is arguably the simplest meaningful problem one could consider. If results and rules are to be found, they ought to make themselves plain in this problem. For certain problem parametrizations (meaning the signals and correlation coefficient), optimal quantization is achievable via a single threshold applied to each observation-the same as under independence. In other cases, one observation is best ignored or is quantized with two thresholds; neither behavior is seen under independence. Further, and again in distinction from the case of independence, it is seen that in certain situations, an XOR fusion rule is optimal, and in these cases, the implied decision rule is bizarre. The analysis is extended to the multivariate Gaussian problem.     

基于PLS的加权朴素贝叶斯分类测试算法

朴素贝叶斯算法是一种简单而高效的分类算法,但是它的条件独立性假设影响了其分类性能。通过放松朴素贝叶斯假设,可以增强其分类效果,但通常会导致计算代价大幅提高。文章提出了一种基于偏最小二乘的加权朴素贝叶斯分类算法,通过建立条件属性和决策属性之间偏最小二乘回归方程,把回归系数赋给对应的条件属性,作为相应的权重,从而在保持简单性的基础上有效地提高了朴素贝叶斯算法的分类性能。最后,通过在UCI数据集上的仿真实验,验证了该算法的有效性。     

A test of independence based on a generalized correlation function

In this paper, we propose a novel test of independence based on the concept of correntropy. We explore correntropy from a statistical perspective and discuss its properties in the context of testing independence. We introduce the novel concept of parametric correntropy and design a test of independence based on it. We further discuss how the proposed test relaxes the assumption of Gaussianity. Finally, we discuss some computational issues related to the proposed method and compare it with state-of-the-art techniques.     

Testing for stochastic independence: application to blind source separation

In this paper, we address the issue of testing for stochastic independence and its application as a guide to selecting the standard independent component analysis (ICA) algorithms when solving blind source separation (BSS) problems. Our investigation focuses on the problem of establishing tests for the quality of separation among recovered sources obtained by ICA algorithms in an unsupervised environment. We review existing tests and propose two contingency table-based algorithms. The first procedure is based on the measure of goodness-of-fit of the observed signals to the model of independence provided by the power-divergence (PD) family of test statistics. We provide conditions that guarantee the validity of the independence test when the individual sources are nonstationary. When the sources exhibit significant time dependence, we show how to adopt Hotelling's T/sup 2/ test statistic for zero mean to create an accurate test of independence. Experimental results obtained from a variety of synthetic and real-life benchmark data sets confirm the success of the PD-based test when the individual source samples preserve the so-called constant cell probability assumption as well as the validity of the T/sup 2/-based test for sources with significant time dependence.     

一种基于维规约的属性加权朴素贝叶斯算法

朴素贝叶斯算法是一种简单而高效的分类算法,但它的属性的条件独立性假设在现实中往往不成立,而且算法本身对高维数据不敏感,如何提高高维数据的分类性能是一个重要的问题.通过确定权重系数进行算法改进,用改进的算法对基于条件信息熵、主成分分析和独立成分分析处理的数据进行分类,并分析性能.     

Mathematical Models for Cochannel Interference in FH/MFSK Multiple-Access Systems

FH/MFSK has been Proposed for a multiuser spread spectrum digital communication system to combat both self-jamming and intentional jamming. An independence assumption at the energy detector outputs is used to evaluate such a system for a digitized voice mobile radio system [11]. This paper presents a correlated model and compares its performance to an independence model under the symbol error probability criterion. A Gaussian process model is also developed for comparison. It is found that the conventional Gaussian approximation is inadequate in predicting the number of users that can be accommodated by an FH/MFSK system using the conventional receiver. Furthermore, for the cases we evaluated, the independence model provides an excellent approximation to the correlated model. Additionally, a conditional Chernoff bound is presented for the more general case of frequency-hopping systems with multiple hops per symbol.     

Acquisition of direct sequence spread spectrum signals withcorrelated fading

Antenna diversity is an effective way of improving the performance of a communication system, especially for initial acquisition in a narrowband code division multiple access (CDMA) system that operates in a flat fading environment. Earlier attempts to analyze such systems assume independent fading at various antenna elements. However, the diversity gain predicted by these analyses is often unrealizable, since the assumption about the independence of the fades is unrealistic in a spatially constrained system. We relax the assumption about the spatial independence of the fades by employing a realistic model for spatial correlation. Also, we obtain an optimum receiver for initial acquisition that takes into account the particulars of the spatial correlation and present a technique for the analysis of this receiver. We also study the diversity gain available for various parameters of the spatial correlation model and conclude that there exists an optimum number of the antenna elements for an initial acquisition system, based on the spatial correlation of the fades. Further, we consider a wideband multicarrier CDMA system with correlated fades among its subcarriers, which also employs antenna diversity, and apply the techniques used in the narrowband case to obtain an optimum receiver and analyze its performance. Finally, we study the tradeoff between the diversity gain available and the noncoherent loss incurred as the numbers of subcarriers and antenna elements are changed     

A New Three-Parameter Extension to the Birnbaum-Saunders Distribution

The Binrbaum-Saunders (B-S) distribution was derived in 1969 as a lifetime model for a specimen subjected to cyclic patterns of stresses and strains, and the ultimate failure of the specimen is assumed to be due to the growth of a dominant crack in the material. The derivation of this model will be revisited, and because the assumption of independence of crack extensions from cycle to cycle can be quite unrealistic, one new model will be derived by relaxing this independence assumption. Here, the sequence of crack extensions is modeled as a long memory process, and characteristics of this development introduces a new third parameter. The model is investigated in detail, and interestingly the original B-S distribution is included as a special case. Inference procedures are also discussed, and an example dataset is used for model comparison.     

Application of coherence theory to propagation of a quasi-monochromatic field in a random medium

Starting from the partial differential equation for scalar wave propagation in a random medium, an integral equation is derived for the propagation of the second-order correlation function. In order to obtain the exact integral equation without resorting to perturbation theory, it is necessary to assume statistical independence of the wave function and the refractive index fluctuations. The physical basis for this assumption is reviewed, and a criterion for its applicability is proposed. The integral equation for the second-order correlation is then transformed into a wavenumber-spectral representation, resulting in appreciable simplification and permitting some direct interpretation.