An experimental model of Chinese textual database

Abstract

A textual database deals with retrieval and manipulation of documents. It allows a user to search on‐line complete documents or parts of documents rather than attributes of documents. Resembling a formatted database which uses a data model as its underlying structure, a textual database has to base its development upon a document model. In this paper, a document model, called the ECHO model, is proposed. The ECHO model provides a document representation, called the ECHO structure, for expressing documents and operations on the representation that serve to express queries and manipulations on documents. It has the ability to provide multiple document structures for a document, a flexible search unit for retrieving textual information, and a subrange search on a textual database. In addition, the ECHO structure is relatively easy to maintain. An architecture of a textual database based on the ECHO model is also proposed. In order to improve the query performance, a refined character inversion method, called ARCIM, is proposed as the text‐access method of the Chinese textual database. The ARCIM can retrieve texts faster than a simple inversion method and requires less space overhead.     

基于双级特征提取与度量的图像检索算法

目的 针对单一低层特征在语义属性中的信息易丢失,导致其对图像描述能力不强,使其检索精度不佳的问题,结合颜色矩（CM）、角径向变换描述符（ART）和边缘直方图（EH）等3种特征,定义一种双级特征提取与度量的图像检索方案。方法 首先,将图像转换为HSV色彩空间,并将其分割为若干个非重叠子图像,通通过计算每个子图像的均值、标准差和偏斜度来表征CM;再利用Euclidean距离,对查询图像和数据库图像的CM进行提取与度量,将输出的检索结果标记为一个图像集。随后,提取查询图像与图像集中每个目标的ART和EH特征;利用Euclidean距离分别度量查询图像与图像集中目标的ART与EH的相似性;最后,对ART与EH的加权组合,输出相似性最高的检索图像。结果 实验表明,与当前常见的检索算法比较,文中算法具有更高的检索精度,表现出更优异的Precision-Recall曲线。结论 所提算法具有良好的检索准确度,在信息处理、包装商标等领域具有一定的参考价值。     

一种IDS报警过滤算法及实现架构研究

为提高入侵检测的有效性,提出了一种基于二级决策进行报警过滤从而消除误报、滥报问题的方法,设计实现了一种基于报警缓冲池的报警优化过滤算法,并对算法进行了效率分析和实验。实验结果表明,该技术可以有效地消除误报、滥报现象,具有较强的实用价值。     

Terrain and Model Queries Using Scalar Representations With Wavelet Compression

In this paper, we present efficient height/distance field data structures for line-of-sight (LOS) queries on terrains and collision queries on arbitrary 3-D models. The data structure uses a pyramid of quad-shaped regions with the original height/distance field at the highest level and an overall minimum/maximum value at the lower levels. The pyramid can compactly be stored in a wavelet-like decomposition but using max and plus operations. Additionally, we show how to get minimum/maximum values for regions in a wavelet decomposition using real algebra. For LOS calculations, we compare with a kd-tree representation containing the maximum height values. Furthermore, we show that the LOS calculation is a special case of a collision detection query. Using our wavelet-like approach, even general and arbitrary collision detection queries can efficiently be answered.      

基于泽尼克矩耦合颜色空间加权度量的商标检索算法

目的 针对商标检索系统中利用单一特征进行识别和度量时,往往难以充分表征商标特征,易出现检索精度和鲁棒性不高等问题,文中拟设计一种泽尼克（Zernike）矩耦合颜色空间加权度量的商标检索方案。方法 首先,利用Zernike矩作为商标的形状描述符,充分描述商标的形状信息。随后,利用颜色空间来描述图像中像素空间信息的颜色分布特征。然后,分别将输入商标的Zernike矩特征、颜色空间特征与存储在数据库中的特征进行匹配,以计算Zernike矩特征的加权Euclidean距离与颜色空间度量。最后,联合颜色空间度量与Euclidean距离,综合考虑形状与颜色特征,形成新的距离测量规则,输出与查询商标相似的商标。结果 实验数据表明,与当前商标检索算法相比较,所提算法具有更高的检索准确率与鲁棒性,表现出更为理想的Precision-Recall以及平均准确率（Mean Average Precision, MAP）。结论 所提算法返回的图像与查询图像相似度较高,在商标注册、侵权保护等方面中具有一定的参考价值。     

MRI brain image classification—a hybrid approach

The present article proposes a novel computer‐aided diagnosis (CAD) technique for the classification of the magnetic resonance brain images. The current method adopt color converted hybrid clustering segmentation algorithm with hybrid feature selection approach based on IGSFFS (Information gain and Sequential Forward Floating Search) and Multi‐Class Support Vector Machine (MC‐SVM) classifier technique to segregate the magnetic resonance brain images into three categories namely normal, benign and malignant. The proposed hybrid evolutionary segmentation algorithm which is the combination of WFF(weighted firefly) and K‐means algorithm called WFF‐K‐means and modified cuckoo search (MCS) and K‐means algorithm called MCS‐K‐means, which can find better cluster partition in brain tumor datasets and also overcome local optima problems in K‐means clustering algorithm. The experimental results show that the performance of the proposed algorithm is better than other algorithms such as PSO‐K‐means, color converted K‐means, FCM and other traditional approaches. The multiple feature set comprises color, texture and shape features derived from the segmented image. These features are then fed into a MC‐SVM classifier with hybrid feature selection algorithm, trained with data labeled by experts, enabling the detection of brain images at high accuracy levels. The performance of the method is evaluated using classification accuracy, sensitivity, specificity, and receiver operating characteristic (ROC) curves. The proposed method provides highest classification accuracy of greater than 98% with high sensitivity and specificity rates of greater than 95% for the proposed diagnostic model and this shows the promise of the approach. © 2015 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 25, 226–244, 2015     

基于双信息统计与引力聚类的图像篡改检测算法

目的 为了解决当前图像复制-粘贴篡改检测算法的鲁棒性与检测精准度不佳等问题。方法 将图像的颜色信息引入伪造检测过程,提出双信息统计机制耦合引力聚类的图像复制-粘贴篡改检测算法。首先,利用Hessian矩阵来准确提取图像的特征点。然后,利用图像的梯度直方图来描述图像的方向特征,并联合图像的颜色信息,构造双信息统计机制,获取图像的特征向量。计算特征向量间的欧氏距离,构造近似测量模型,对图像特征进行匹配。最后,利用引力聚类方法,实现图像特征点的聚类,精准检测复制-粘贴篡改内容。结果 与当前图像复制-粘贴篡改检测方法相比,所提算法具有更高的检测精准度,以及更好的鲁棒性。结论 所提方案可以准确检测并定位出伪造内容,在图像水印、信息安全领域具有一定的参考价值。     

A variable-parameter unsupervised learning clustering neural network approach with application to machine-part group formation

In this paper, we develop a new unsupervised learning clustering neural network method for clustering problems in general and for solving machine-part group formation problems in particular. We show that our new approach solves a very challenging problem in the area of machine-part group formation. A review of machine-part group formation methods and unsupervised learning artificial neural network methods is given. We modify the well-known competitive learning algorithm by using the generalized Euclidean distance, and a momentum term in the weight vector updating equations. The cluster structure can be adjusted by changing the coefficients in the generalized Euclidean distance. The algorithm is flexible and applicable to many practical problems. We also develop a neural network clustering system which can be used to cluster a 0-1 matrix into diagonal blocks. The developed neural network clustering system is independent of the initial matrix and gives clear final clustering results which specify the machines and parts in each group. We use the developed neural network clustering system to solve several machine-part group formation problems, in which the machine-part incidence matrix is to be clustered into a diagonal block structure. An algorithm is developed to consider lower and upper bounds on the number of machines for each cell. The computational results are compared with those from the well-known rank order clustering and directive clustering analysis methods.     

Algorithm for training the minimum error one-class classifier of images

We propose a training algorithm for one-class classifiers in order to minimize the classification error. The aim is to choose the optimal value of the slack parameter, which controls the selectiveness of a classifier. The one-class classifier based on the coordinated clusters representation of images is trained and then used for the classification of texture images. As the slack parameter C varies through a range of values, for each C, the misclassification rate is computed using only the training samples. The value of C that yields the minimum misclassification rate, estimated over the training set, is taken as the optimal value, C(opt). Finally, the optimized classifier is tested on the extended database of images. Experimental results demonstrate the validity of the proposed method. In our experiments, classification efficiency approaches, or is equal to, 100%, after the optimal training of the classifier.     

基于区域生长耦合多分类器的商标检索算法

目的针对商标检索算法中易出现的语义鸿沟,底层视觉特征与高层语义相关性不强而导致商标检索精度不理想的问题,定义一种基于区域生长耦合多分类器的商标检索方案。方法首先对输入的商标进行预处理,去除图像中的噪声和杂散点,并通过3D直方图和聚类算法来提取输入图像中的主颜色;基于区域生长算法,合并具有相同颜色标签的所有连接点,以形成颜色区域;然后根据生成的颜色区域,分别定义颜色分类器、形状分类器和关系分类器,利用每个分类器计算查询图像和数据库中图像的检索优势概率;最后通过决策组合,根据检索规则和列表长度找到最相似的商标,并利用动态选择方案进一步提高检索准确率。结果实验结果表明,与当前商标检索方案相比,所提检索系统具有更为理想的Precision-Recall曲线,对缩放、扭曲和噪声具有更高的鲁棒性。结论所提方案在各类几何变换下具备较高的检索准确率,对商标注册、版权保护等行业有较好的借鉴意义。     

An Intrusion Detection Algorithm Based on Feature Graph

With the development of Information technology and the popularization of Internet, whenever and wherever possible, people can connect to the Internet optionally. Meanwhile, the security of network traffic is threatened by various of online malicious behaviors. The aim of an intrusion detection system (IDS) is to detect the network behaviors which are diverse and malicious. Since a conventional firewall cannot detect most of the malicious behaviors, such as malicious network traffic or computer abuse, some advanced learning methods are introduced and integrated with intrusion detection approaches in order to improve the performance of detection approaches. However, there are very few related studies focusing on both the effective detection for attacks and the representation for malicious behaviors with graph. In this paper, a novel intrusion detection approach IDBFG (Intrusion Detection Based on Feature Graph) is proposed which first filters normal connections with grid partitions, and then records the patterns of various attacks with a novel graph structure, and the behaviors in accordance with the patterns in graph are detected as intrusion behaviors. The experimental results on KDD-Cup 99 dataset show that IDBFG performs better than SVM (Supprot Vector Machines) and Decision Tree which are trained and tested in original feature space in terms of detection rates, false alarm rates and run time.     

Computational intelligence techniques for human brain MRI classification

This study proposes an image classification methodology that automatically classifies human brain magnetic resonance (MR) images. The proposed methods contain four main stages: Data acquisition, preprocessing, feature extraction, feature reduction and classification, followed by evaluation. First stage starts by collecting MRI images from Harvard and our constructed Egyptian database. Second stage starts with noise reduction in MR images. Third stage obtains the features related to MRI images, using stationary wavelet transformation. In the fourth stage, the features of MR images have been reduced using principles of component analysis and kernel linear discriminator analysis (KLDA) to the more essential features. In last stage, the classification stage, two classifiers have been developed to classify subjects as normal or abnormal MRI human images. The first classifier is based on K‐Nearest Neighbor (KNN) on Euclidean distance. The second classifier is based on Levenberg‐Marquardt (LM‐ANN). Classification accuracy of 100% for KNN and LM‐ANN classifiers has been obtained. The result shows that the proposed methodologies are robust and effective compared with other recent works.     

An Entity-Association-Based Matrix Factorization Recommendation Algorithm

Collaborative filtering is the most popular approach when building recommender systems, but the large scale and sparse data of the user-item matrix seriously affect the recommendation results. Recent research shows the user’s social relations information can improve the quality of recommendation. However, most of the current social recommendation algorithms only consider the user's direct social relations, while ignoring potential users’ interest preference and group clustering information. Moreover, project attribute is also important in item rating. We propose a recommendation algorithm which using matrix factorization technology to fuse user information and project information together. We first detect the community structure using overlapping community discovery algorithm, and mine the clustering information of user interest preference by a fuzzy clustering algorithm based on the project category information. On the other hand, we use project-category attribution matrix and user-project score matrix to get project comprehensive similarity and compute project feature matrix based on Entity Relation Decomposition. Fusing the user clustering information and project information together, we get Entity-Association-based Matrix Factorization (EAMF) model which can be used to predict user ratings. The proposed algorithm is compared with other algorithms on the Yelp dataset. Experimental studies show that the proposed algorithm leads to a substantial increase in recommendation accuracy on Yelp data set.     

An Energy-Efficient Protocol Using an Objective Function & Random Search with Jumps for WSN

Wireless Sensor Networks (WSNs) have hardware and software limitations and are deployed in hostile environments. The problem of energy consumption in WSNs has become a very important axis of research. To obtain good performance in terms of the network lifetime, several routing protocols have been proposed in the literature. Hierarchical routing is considered to be the most favorable approach in terms of energy efficiency. It is based on the concept parent-child hierarchy where the child nodes forward their messages to their parent, and then the parent node forwards them, directly or via other parent nodes, to the base station (sink). In this paper, we present a new Energy-Efficient clustering protocol for WSNs using an Objective Function and Random Search with Jumps (EEOFRSJ) in order to reduce sensor energy consumption. First, the objective function is used to find an optimal cluster formation taking into account the ratio of the mean Euclidean distance of the nodes to their associated cluster heads (CH) and their residual energy. Then, we find the best path to transmit data from the CHs nodes to the base station (BS) using a random search with jumps. We simulated our proposed approach compared with the Energy-Efficient in WSNs using Fuzzy C-Means clustering (EEFCM) protocol using Matlab Simulink. Simulation results have shown that our proposed protocol excels regarding energy consumption, resulting in network lifetime extension.     

基于模糊K-均值算法的模糊分类器设计

基于模糊K-均值算法的模糊分类器,就是把目前比较常用的模糊K-均值算法的聚类方法,再一次与模糊分类规则提取相结合而得到的一种分类器。它是一种很有效的模糊分类器,训练样本能正确的分类。在这种方法中,首先用模糊K-均值算法按剖分和覆盖的原则把训练样本分成群,并且每一群的中心和半径都被计算出来。然后,设计一个用模糊规则来表示分类的模糊系统。这样就有效地构建了一个能对训练样本比较准确分类的模糊分类器。用这种方法设计的分类器不需要预定义参数、训练时间较短、方法简单     

基于空间邻域信息的二维模糊聚类图像分割

传统模糊C均值聚类(FCM)算法进行图像分割时仅利用了像素的灰度信息,并且使用对噪声较敏感的欧氏距离作为像素与聚类中心距离度量的标准,因此抗噪性能较差.为了克服传统FCM算法的局限性,本文提出了一种基于空间邻域信息的二维模糊聚类图像分割方法(2DFCM).该方法利用二维直方图描述的像素邻域关系属性,一方面为聚类提供较准确的初始聚类中心,从而避免聚类中的死点问题;另一方面通过提出聚类中心同时在像素值、像素邻域值二维方向上进行更新的思想,建立了包含邻域信息的新的聚类目标函数,实现了图像的分割.实验结果表明,这种方法抗噪能力强、收敛速度快,是一种有效的模糊聚类图像分割方法.     

Fuzzy relevance vector machine based classification of lung nodules in computed tomography images

Lung cancer is a critical disease with growing death rate, hence, the faster identification and treatment of lung cancer is essential. In medical image processing, the traditional methods like support vector machine, relevance vector machine for classifying cancer tissues are less sensitive to false data and required optimal improvement in classification accuracy. The proposed system of accurate lung cancer classification is obtained by a hybrid fuzzy relevance vector machine (FRVM) classifier with correlation negation ant colony optimization (CNACO) algorithm. This system provides enhanced accuracy and sensitivity by implementing two stages of feature extraction, image thresholding, and tumor segmentation, with a novel feature selection and tumor classification algorithm. The best features are selected by the proposed CNACO algorithm. The selected features are labeled and classified by FRVM classifier. The proposed classification scheme is validated on lung image database consortium and image database resource initiative public database and obtained accuracy of about 98.75%.     

Enhanced Deep Autoencoder Based Feature Representation Learning for Intelligent Intrusion Detection System

In the era of Big data, learning discriminant feature representation from network traffic is identified has as an invariably essential task for improving the detection ability of an intrusion detection system (IDS). Owing to the lack of accurately labeled network traffic data, many unsupervised feature representation learning models have been proposed with state-of-the-art performance. Yet, these models fail to consider the classification error while learning the feature representation. Intuitively, the learnt feature representation may degrade the performance of the classification task. For the first time in the field of intrusion detection, this paper proposes an unsupervised IDS model leveraging the benefits of deep autoencoder (DAE) for learning the robust feature representation and one-class support vector machine (OCSVM) for finding the more compact decision hyperplane for intrusion detection. Specially, the proposed model defines a new unified objective function to minimize the reconstruction and classification error simultaneously. This unique contribution not only enables the model to support joint learning for feature representation and classifier training but also guides to learn the robust feature representation which can improve the discrimination ability of the classifier for intrusion detection. Three set of evaluation experiments are conducted to demonstrate the potential of the proposed model. First, the ablation evaluation on benchmark dataset, NSL-KDD validates the design decision of the proposed model. Next, the performance evaluation on recent intrusion dataset, UNSW-NB15 signifies the stable performance of the proposed model. Finally, the comparative evaluation verifies the efficacy of the proposed model against recently published state-of-the-art methods.     

面向海量NetFlow数据的存储和查询处理方法研究

针对全国骨干网高速海量Net Flow数据到来速度快、数据量大以及对所存数据进行频繁多维查询操作的特点,提出了一种多维属性聚簇存储(MACS)模型。该模型根据实际应用环境中查询的特点对数据进行空间分片,以并行加流水的方式对数据进行存储。此外,为Net Flow提出了一种超多面体的查询模式。真实环境实验结果表明,运用MACS模型实现的系统单点数据实时存储速度达到270万条/s,远远快于其他的数据分析系统,并且多维属性查询的速度优于Hive和Impala。