An architectural approach to minimizing feature interactions intelecommunications

Feature interaction, in the context of telecommunications, is the phenomenon where a user (end-user or system operator) observes services failing to perform as expected, and where the failures are due to the presence of other services (or multiple instances of a single service) in the network. We propose an architectural approach to minimize the problem of feature interaction while increasing the reusability of software for services. Our approach recognizes that currently many assumptions about service operations and system capabilities are implicitly built into the service software architecture. We claim that many nontrivial feature interactions arise when services or technologies with new capabilities, based on conflicting assumptions, are introduced. We propose an architecture based on software agents and separation of concerns. A processing model for service-independent interaction management within the agent architecture is discussed. This architecture effectively removes many common assumptions about roles and technology from the service software architecture, thus reducing the need to provide ad hoc solutions to particular feature interactions. We demonstrate the power of such an architecture by showing how it deals with examples of interactions among existing services, interactions introduced by Universal Personal Telecommunications (UPT), and its applicability to emerging broadband and multimedia service requirements. We also consider the architecture in the context of the intelligent network     

An enhanced feature selection filter for classification of microarray cancer data

The main aim of this study is to select the optimal set of genes from microarray cancer datasets that contribute to the prediction of specific cancer types. This study proposes the enhancement of the feature selection filter algorithm based on Joe's normalized mutual information and its use for gene selection. The proposed algorithm is implemented and evaluated on seven benchmark microarray cancer datasets, namely, central nervous system, leukemia (binary), leukemia (3 class), leukemia (4 class), lymphoma, mixed lineage leukemia, and small round blue cell tumor, using five well‐known classifiers, including the naive Bayes, radial basis function network, instance‐based classifier, decision‐based table, and decision tree. An average increase in the prediction accuracy of 5.1% is observed on all seven datasets averaged over all five classifiers. The average reduction in training time is 2.86 seconds. The performance of the proposed method is also compared with those of three other popular mutual information–based feature selection filters, namely, information gain, gain ratio, and symmetric uncertainty. The results are impressive when all five classifiers are used on all the datasets.     

Detection and resolution of feature interactions in IP multimedia subsystem

The IP Multimedia Subsystem (IMS) enables the delivery of multimedia services through different access networks. 3GPP adopted the Session Initiation Protocol (SIP) as the IMS signalling protocol. However, IMS interaction with these services faces the ‘feature interaction’ problem encountered in traditional telephony networks. This problem occurs as features invoked during a session behave correctly when processed separately from each other, but not when running together. Although much research work has explored managing feature interactions, applying these solutions in IMS remains challenging and further investigation is needed. This paper aims to analyze the feature interaction issue in IMS and to propose a new solution. We defined a SIP‐based algorithm and the associated mechanisms to enhance IMS service invocation. This algorithm is the core of the Service Broker, a new functional entity in charge of managing feature interaction. We validated our proposal through a performance evaluation and prototyped the Service Broker on an open source IMS platform. Copyright © 2008 John Wiley & Sons, Ltd.     

Gabor feature based classification using the enhanced fisher lineardiscriminant model for face recognition

This paper introduces a novel Gabor-Fisher (1936) classifier (GFC) for face recognition. The GFC method, which is robust to changes in illumination and facial expression, applies the enhanced Fisher linear discriminant model (EFM) to an augmented Gabor feature vector derived from the Gabor wavelet representation of face images. The novelty of this paper comes from (1) the derivation of an augmented Gabor feature vector, whose dimensionality is further reduced using the EFM by considering both data compression and recognition (generalization) performance; (2) the development of a Gabor-Fisher classifier for multi-class problems; and (3) extensive performance evaluation studies. In particular, we performed comparative studies of different similarity measures applied to various classifiers. We also performed comparative experimental studies of various face recognition schemes, including our novel GFC method, the Gabor wavelet method, the eigenfaces method, the Fisherfaces method, the EFM method, the combination of Gabor and the eigenfaces method, and the combination of Gabor and the Fisherfaces method. The feasibility of the new GFC method has been successfully tested on face recognition using 600 FERET frontal face images corresponding to 200 subjects, which were acquired under variable illumination and facial expressions. The novel GFC method achieves 100% accuracy on face recognition using only 62 features.     

融合颜色特征和纹理特征的图像检索算法

文章结合颜色特征和纹理特征,利用欧氏距离计算两幅图像的相似度进行图像检索。首先,在HSV空间下,提取颜色特征并进行归一化处理。然后,提取GLCM的特征值,并结合Tamura特征形成更加丰富的纹理特征。最后,分别计算待检索图像和图像库中图像的颜色和纹理相似度,在不同权重下融合颜色和纹理特征得出最终相似度。Matlab实验表明,不同种类的图像,分配颜色和纹理的不同权重时,有不同的查准率。调整图像的特征权重可以提高查准率。     

Nonparametric feature selection

Two groups ofL-dimensional observations of sizeN_{1}andN_{2}are known to be random vector variables from two unknown probability distribution functions [1]. A method is discussed for obtaining anl-dimensional linear subspace of the observation space in which thel-variate marginal distributions are most separated, based on a nonparametric estimate of probability density functions and a distance criterion. The distance used essentially is theL_{2}norm of the difference between Parzen estimates of the two densities. An algorithm is developed that determines the subspace for which the distance between the two densities is maximized. Computer simulations are performed.     

局部和全局特征融合的红外图像显著性检测

为了更好地凸显复杂环境的红外目标特征,提出 一种融合局部和全局特征的红外图像 显著性检测方法。在获取图像超像素的基础上,提取每个区域空间距离加权的邻域对比度特 征,并考虑区域大小和位置的影响,构建局部显著图;然后提取每个区域空间距离加权的全 局灰度特征,构建全局显著图;最后融合局部和全局显著图,实现图像显著性检测。实验结 果 表明,本文方法的显著图结果目标区域一致高亮且边缘清晰,同时背景杂波抑制效果好。无 论 主观评价还是客观指标,本文方法都优于当前流行的图像显著性检测方法。     

非合作目标局部特征识别轻量化特征融合网络设计

给出一种基于轻量化卷积神经网络的空间非合作目标局部特征检测网络,即NCDN模型。在SSD模型中引入特征融合策略以适应不同距离下的检测需求,提高模型对图像尺度变换引起局部特征分辨率降低的鲁棒性;并采用不同压缩比例对MobileNetV2内部卷积通道数量做压缩,从而得到轻量化特征提取网络;对SPEED数据集进行局部特征标注与训练以验证NCDN适用的距离范围。实验结果表明,该模型能够在45 m内距离范围保证mAP达到0.90,同时通道压缩节省75%计算量后模型精度损失仅为5%。满足在轨检测精度和计算量需求。     

基于小波变换的网络流量异常监测仿真实验

为有效定位识别和提取网络流量序列的暂态性异常特征,针对网络异常流量特征扰动性和暂态性特点,提出一种基于小波分解的二叉分类回归决策树主分量特征优化跟踪特征提取算法。利用训练集建立决策树模型,采用二叉分类回归决策树模型进行主分量特征优化跟踪建模,利用双正交提升小波分解得到的各层细节信号对暂态性扰动特征的敏感性,通过小波分解得到各层细节信号,将提取的小波分层细节信号的奇异值分解特征再返回到决策树主分量特征优化跟踪模型中,实现网络流量异常特征的定位提取和识别。仿真实验表明,改进算法的抗干扰能力和分辨率提高显著,暂态性异常特征谱图分辨能力提高,异常特征分布谱清晰可见,展示了较好的特征提取和状态识别性能。     

卫星的红外辐射特征研究

卫星的红外辐射特征对于工作卫星与失效卫星的判别具有重要意义,根据卫星与地球、太阳之间的位置关系,计算了太阳直接辐射角系数,利用随机模拟法计算了地球反照和地球红外辐射角系数,建立了卫星温度控制方程,并利用辐射传递系数的概念和蒙特卡洛方法进行辐射换热的计算,对卫星温度场进行了求解,讨论了卫星红外辐射特征的计算方法,并对其特性进行了分析计算,结果表明：卫星的散热面是区别工作卫星和失效卫星的依据。     

An OpenCL-based feature matcher

Nowadays visual search is one of the most active branches of computer vision. It relies on finding invariant points inside images, describing them into features and then matching these features against a reference database to identify objects in the scene or the entire photo (environment). In this paper, we discuss an approach to feature matching that exploits the capabilities of modern GPUs to speed up the aforementioned and that keeps low the number of false matches.     

基于语谱图提取深度空间注意特征的语音情感识别算法

从语音情感特征的提取和分类建模出发,以混合卷积神经网络模型为基础,改进特征提取中的 Itti模型,包括：增加通过局部二值模式提取的纹理特征;结合听觉敏感度权重提取情感强相关特征。然后提出通过特征约束条件提取标定权重特征的约束挤压和激励网络结构;最后形成以 VGGnet 和长短时记忆网络混合网络为基础的微调模型,进一步提升了情感表征能力。通过在自然情感数据库和柏林德语数据库上进行验证,该模型在情感识别率上有明显的上升,相较于基准模型提升了 8. 43%,同时对比了本模型在自然数据库（FAU-AEC）和柏林数据库（EMO-DB）上的识别效果,实验结果证明模型具有良好的泛化性。     

Joint feature weighting and adaptive graph-based matrix regression for image supervised feature Selection

Matrix regression (MR) is a regression model that can directly perform on matrix data. However, the effect of each element in matrix data on regression model is different. Taking into consideration the relevance of every original feature in the matrix data and their influence on the final estimation of the regression model, we introduce an unknown weight matrix to encode the relevance of feature in matrix data and propose a feature weighting and graph-based matrix regression (FWGMR) model for image supervised feature selection. In this model, the feature weight matrix is used to select some important features from the matrix data and preserve the relative spatial location relationship of elements in the matrix data. In addition, in order to effectively and reasonably preserve the local manifold structure of the training matrix samples, a regularization term in the model is used to adaptively learn a graph matrix on low-dimensional space. An optimization algorithm is devised to solve FWGMR model and to provide the closed-form solutions of this model in each iteration. Extensive experiments on some public datasets demonstrate the superiority of FWGMR.     

基于改进增强特征选择算法的特征融合图像隐写分析

针对现有的基于特征融合的JPEG隐写分析方法特征冗余度高、通用性较低的问题,提出了一种基于改进的增强特征选择(BFS,boosting feature selection)算法的通用JPEG隐写分析方法。从线性相关度和非线性相关度两方面降低特征冗余,将特征自相关系数和互信息这两种统计性能引入到特征的评价准则中,重新设计了特征权重计算方法,改进了BFS算法的特征评价函数。通过改进的BFS特征选择算法将3组互补性较强且准确率高的特征进行融合降维,得到最优特征子集训练分类器。对3种高隐蔽性隐写算法F5、Outguess和MME3,在不同嵌入率下进行了大量实验。结果表明,本文方法的分析准确率高于现有的检测率较高的JPEG隐写分析方法和典型的融合分析方法,融合后的特征相关性明显下降,并且具有更强的通用性。     

数字化变电站的特点浅析

随着智能化电气设备的发展,数字化变电站的建设已成为可能。文章主要从数字化变电站的一次设备、二次设备及网络结构等几个方面进行论述,指出了数字化变电站的发展趋势。     

Contrast enhanced photolithography

This paper introduces a new method, contrast enhancement, which improves the imaging properties of conventional photoresist through the use of a photobleachable layer applied to the resist surface. Contrast enhancement allows high contrast photopatterns to be formed even when low contrast illumination is used. A short discussion of the theory and an experimental demonstration of the method are presented.     

特征选择算法综述

自20世纪90年代以来,特征选择成为模式识别和机器学习领域的重要研究方向,研究成果十分显著,但是也存在许多问题需要进一步研究。本文首先将特征选择视为特征集合空间中的启发式搜索问题,对特征选择涉及的四个要素进行了阐述,然后从各个角度对特征选择算法进行了分类,概述了其各个分支的发展态势,最后探讨了基于多目标免疫优化的特征选择方法的研究思路。     

Resolution enhanced ESPRIT

The authors present a variant of ESPRIT, known as the resolution enhanced (RE)-ESPRIT, that lowers resolution thresholds by incorporating left weighting of the signal eigenvectors to provide sector focusing. They discuss the advantages of this new left weighted ESPRIT. A compact expression for the asymptotic covariance matrix of the estimates is presented. Analytical and simulation results are included to show the resolution improvement of ESPRIT