Search

In order to improve the performance of image segmentation, this paper presented a gray level jump segmentation algorithm, which defined the direction of the texture, simultaneously, calculated the width of ridge line, gave the distance characteristics between textures, and established the mathematical model of the texture border, accordingly presented a new texture segmentation algorithm and compared with other texture segmentation algorithms. The simulation results show that the segmentation algorithm has some advantages to texture segmentation, such as has higher segmentation precision, faster segmentation speed, stronger anti-noise capability, less lost information of target, and so on. The segmented regions hardly contain other texture regions and background region. Moreover, this paper extracted the characteristic points and characteristic parameters in various segmented regions for texture image to obtain the characteristic vector, compared the characteristic vector with the standard template vectors, and identified the type of target in a range of threshold value. Experimental results show that the proposed target recognition approach has higher recognition rate and faster recognition speed than the existing target recognition approaches. Advancements in image processing through the study of texture segmentation are not only applicable to image fields, but also are of important theoretical value to target recognition. These researches in this paper will play an important role in a theoretical reference and practical significance to the development of all target recognition departments based on image system such as the aerospace, public security, road traffic, and so on. [ABSTRACT FROM AUTHOR]

Highlights • The proposed method hides sensitive rules using electromagnetic field optimization. • Several sensitive association rules are hided simultaneously in the method. • The proposed method also has fewer lost rules than other well-known algorithms. • Two fitness functions are proposed to find the solution with minimum side effects. • The method is evaluated on both real-world and synthetic datasets. Abstract Privacy preserving data mining has been a major research subject in recent years. The most important goal of this area is to protect personal information and prevent disclosure of this information during the data mining process. There are various techniques in the field of privacy preserving data mining. One of these techniques is association rules mining. The main purpose of association rules mining is to hide sensitive association rules. So far, various algorithms have been presented to this field in order to reach the purpose of sensitive association rules hiding. Each algorithm has its own specific functions and methods. To hide sensitive association rules, this paper presents an electromagnetic field optimization algorithm (EFO4ARH). This algorithm utilizes the data distortion technique to hide the sensitive association rules. In this algorithm, two fitness functions are used to reach the solution with the least side effects. Also, in this algorithm, the runtime has been reduced. This algorithm consists of a technique for exiting from local optima point and moving toward global optimal points. The performance of the proposed algorithm is evaluated by doing experiments on both real-world and synthetic datasets. Compared to four reference algorithms, the proposed algorithm shows a reduction in the side effects and better preservation of data quality. The performance of EFO4ARH is tested by standard deviation and mean Friedman ranks of error for standard functions (CEC benchmarks). In addition, hiding experiments show that our proposed algorithm outperforms existing hiding algorithms. [ABSTRACT FROM AUTHOR]

Traditional document-level sentiment analysis fails to consider sentiment sequence within documents. This research paper proposes a novel perspective of sequence-based document sentiment analysis for discovering sentiment sequence and clustering sentiments for Email data. The proposed scheme of approach applies a trajectory clustering algorithm to Email trajectories transformed from sentiment features generated from SentiWordNet lexicon for discovering sentiment sequence within topic and temporal pattern distributions on the basis of trajectory clusters and their representative trajectories. Experiments conducted on real Email data provide evidence on proving the feasibility of the proposed technique and justifying the indispensability of sentiment sequence within documents in the determination of sentiment polarity. [ABSTRACT FROM AUTHOR]

The discovery of community structure is the basis of understanding the topology structure and social function of the network. It is also an important factor for recommendation technology, information dissemination, event prediction, and more. In this paper, we consider the structure and characteristics of the social network and propose an algorithm based on semi-supervised matrix factorization and random walk. The proposed method first calculates the transition probability between nodes through the topology of the network. The random walk model is then used to obtain the final walk probability, and the feature matrix is constructed. At the same time, we combine a priori content information in the network to build a must-link matrix and a cannot-link matrix. We then merge them into the feature matrix of the random walk to form a new feature matrix. Finally, the expectation of the number of edges is defined according to the factorized membership matrix. Results demonstrate the effectiveness and better performance of our method. [ABSTRACT FROM AUTHOR]

The P 300 is an endogenous event-related potential (ERP) that is naturally elicited by rare and significant stimuli, arisen from the frontal, temporal and occipital lobe of the brain, although is usually measured in the parietal lobe. P 300 signals are increasingly used in brain-computer interfaces (BCI) because the users of ERP-based BCIs need no special training. In order to detect the P 300 signal, most studies in the field have been focused on a supervised approach, dealing with over-fitting filters and the need for later validation. In this paper we start bridging this gap by modeling an unsupervised classifier of the P 300 presence based on a weighted score. This is carried out through the use of matched filters that weight events that are likely to represent the P 300 wave. The optimal weights are determined through a study of the data’s features. The combination of different artifact cancelation methods and the P 300 extraction techniques provides a marked, statistically significant, improvement in accuracy at the level of the top-performing algorithms for a supervised approach presented in the literature to date. This innovation brings a notable impact in ERP-based communicators, appointing to the development of a faster and more reliable BCI technology. [ABSTRACT FROM AUTHOR]

In this paper, we present a monocular algorithm for detection of moving objects in roundabouts, developed to be used in Advanced Driver Assistance Systems (ADAS). The proposed algorithm fuses motion-based approach and appearance-based approach to overcome monocular system restrictions in object detection. It processes images sequence which is extracted from video stream of a monocular camera installed on the front left fender, close to the head lamp. The algorithm is customized based on the characteristics of traffic movement in roundabouts and it helps to have a safe entry into the roundabouts. First, the algorithm detects the independent moving features of the images and clusters them to find the moving objects. Then, it uses Haar-like features along side Soft-Cascade AdaBoost to detect the vehicles which are moving towards/away to/from the car. The solution has been tested on some validation set, and also it is used in some real drives. It is shown that the developed algorithm has an acceptable performance in roundabouts. The corresponding results and statistics are reported. [ABSTRACT FROM AUTHOR]

In this paper, we introduce EXPress closED ITemset Enumeration (E xpedite ), a new frequent closed itemset (FCI) miner designed to speed up the process of FCIs extraction from a dataset of transactions. Compared to the state of the art, E xpedite provides a CPU time saving of up to two orders of magnitude without compromising other dimensions of performance (e.g. memory). The reason why it is so fast is that E xpedite wastes less time in mining intermediate item sets that are discarded in later phases of the algorithm. More specifically, it cuts down the number of both duplicate FCIs—those generated multiple times by the algorithm—and infrequent itemsets—those with low support or no supporting transactions. This feature, enjoyable by both sparse and dense datasets, is analytically motivated first, and then experimentally supported by extensive tests on real datasets. As a further contribution, we propose two alternative implementations of E xpedite that perform even better than the basic version, although they rely on particular features of the input dataset. [ABSTRACT FROM AUTHOR]

Ant Colony Extended (ACE) is a novel algorithm belonging to the general Ant Colony Optimisation (ACO) framework. Two specific features of ACE are: the division of tasks between two kinds of ants, namely patrollers and foragers, and the implementation of a regulation policy to control the number of each kind of ant during the searching process. In addition, ACE does not employ the construction graph usually employed by classical ACO algorithms. Instead, the search is performed using a state space exploration approach. This paper studies the performance of ACE in the context of the Travelling Salesman Problem (TSP), a classical combinatorial optimisation problem. The results are compared with the results of two well known ACO algorithms: ACS and MMAS. ACE shows better performance than ACS and MMAS in almost every TSP tested instance. [ABSTRACT FROM AUTHOR]

Abstract: Template matching (TM) plays an important role in several image processing applications such as feature tracking, object recognition, stereo matching and remote sensing. The TM approach seeks the best possible resemblance between a sub-image, known as template, and its coincident region within a source image. TM has two critical aspects: similarity measurement and search strategy. The simplest available TM method finds the best possible coincidence between the images through an exhaustive computation of the Normalized Cross-Correlation (NCC) value (similarity measurement) for all elements in the source image (search strategy). Unfortunately, the use of such approach is strongly restricted since the NCC evaluation is a computationally expensive operation. Recently, several TM algorithms that are based on evolutionary approaches, have been proposed to reduce the number of NCC operations by calculating only a subset of search locations. In this paper, a new algorithm based on the states of matter phenomenon is proposed to reduce the number of search locations in the TM process. In the proposed approach, individuals emulate molecules that experiment state transitions which represent different exploration–exploitation levels. In the algorithm, the computation of search locations is drastically reduced by incorporating a fitness calculation strategy which indicates when it is feasible to calculate or to only estimate the NCC value for new search locations. Conducted simulations show that the proposed method achieves the best balance in comparison to other TM algorithms considering the estimation accuracy and the computational cost. [Copyright &y& Elsevier]

Abstract: In this paper a further generalization of differential evolution based data classification method is proposed, demonstrated and initially evaluated. The differential evolution classifier is a nearest prototype vector based classifier that applies a global optimization algorithm, differential evolution, for determining the optimal values for all free parameters of the classifier model during the training phase of the classifier. The earlier version of differential evolution classifier that applied individually optimized distance measure for each new data set to be classified is generalized here so, that instead of optimizing a single distance measure for the given data set, we take a further step by proposing an approach where distance measures are optimized individually for each feature of the data set to be classified. In particular, distance measures for each feature are selected optimally from a predefined pool of alternative distance measures. The optimal distance measures are determined by differential evolution algorithm, which is also determining the optimal values for all free parameters of the selected distance measures in parallel. After determining the optimal distance measures for each feature together with their optimal parameters, we combine all featurewisely determined distance measures to form a single total distance measure, that is to be applied for the final classification decisions. The actual classification process is still based on the nearest prototype vector principle; A sample belongs to the class represented by the nearest prototype vector when measured with the above referred optimized total distance measure. During the training process the differential evolution algorithm determines optimally the class vectors, selects optimal distance metrics for each data feature, and determines the optimal values for the free parameters of each selected distance measure. Based on experimental results with nine well known classification benchmark data sets, the proposed approach yield a statistically significant improvement to the classification accuracy of differential evolution classifier. [Copyright &y& Elsevier]

Abstract: One of the main problems in pattern recognition is obtaining the best set of features to represent the data. In recent years, several feature extraction algorithms have been proposed. However, due to the high degree of variability of the patterns, it is difficult to design a single representation that can capture the complex structure of the data. One possible solution to this problem is to use a multiple-classifier system (MCS) based on multiple feature representations. Unfortunately, still missing in the literature is a methodology for comparing and selecting feature extraction techniques based on the dissimilarity of the feature representations. In this paper, we propose a framework based on dissimilarity metrics and the intersection of errors, in order to analyze the relationships among feature representations. Each representation is used to train a classifier, and the results are compared by means of a dissimilarity metric. Then, with the aid of Multidimensional Scaling, visual representations are obtained of each of the dissimilarities and used as a guide to identify those that are either complementary or redundant. We applied the proposed framework to the problem of handwritten character and digit recognition. The analysis is followed by the use of an MCS built on the assumption that combining dissimilar feature representations can greatly improve the performance of the system. Experimental results demonstrate that a significant improvement in classification accuracy is achieved due to the complementary nature of the representations. Moreover, the proposed MCS obtained the best results to date for both the MNIST handwritten digit dataset and the Cursive Character Challenge (C-Cube) dataset. [Copyright &y& Elsevier]

Abstract: In sentiment analysis, a finer-grained opinion mining method not only focuses on the view of the product itself, but also focuses on product features, which can be a component or attribute of the product. Previous related research mainly relied on explicit features but ignored implicit features. However, the implicit features, which are implied by some words or phrases, are so significant that they can express the users’ opinion and help us to better understand the users’ comments. It is a big challenge to detect these implicit features in Chinese product reviews, due to the complexity of Chinese. This paper is mainly centered on implicit features identification in Chinese product reviews. A novel hybrid association rule mining method is proposed for this task. The core idea of this approach is mining as many association rules as possible via several complementary algorithms. Firstly, we extract candidate feature indicators based word segmentation, part-of-speech (POS) tagging and feature clustering, then compute the co-occurrence degree between the candidate feature indicators and the feature words using five collocation extraction algorithms. Each indicator and the corresponding feature word constitute a rule (feature indicator → feature word). The best rules in five different rule sets are chosen as the basic rules. Next, three methods are proposed to mine some possible reasonable rules from the lower co-occurrence feature indicators and non indicator words. Finally, the latest rules are used to identify implicit features and the results are compared with the previous. Experiment results demonstrate that our proposed approach is competent at the task, especially via using several expanding methods. The recall is effectively improved, suggesting that the shortcomings of the basic rules have been overcome to certain extent. Besides those high co-occurrence degree indicators, the final rules also contain uncommon rules. [Copyright &y& Elsevier]

Abstract: A federation of data warehouses is understood as a set of data warehouses, which can be processed as a whole in the logic level. Physically, the federation does not gather data into one place. This paper presents a formal framework for data and knowledge processing in data warehouse federations. The management system for a data warehouse federation consists of an user interface enabling presentation of user queries, a program for query decomposition and a program for integrating knowledge coming from different data warehouses as the answers to a user query. We propose a model for query decomposition process and knowledge integration. It contains also the algorithm for knowledge inconstancy processing. This kind of inconsistency often occurs since very often the knowledge extracted from different data warehouses refers to the same subject, but is not consistent. [Copyright &y& Elsevier]

Abstract: Effective data mining solutions have for long been anticipated in Customer Relationship Management (CRM) to accurately predict customer behavior, but from various industry research and case studies we have observed sub-optimal CRM classification models due to inferior data quality inherent to CRM data set. In this paper, one type of CRM data with a distinctive distribution pattern of Reduced Dimensionality is discussed. A new classification framework termed Partial Focus Feature Reduction is proposed to resolve CRM data set with Reduced Dimensionality using a collection of efficient data mining techniques characterizing a specially tailored modality grouping method to significantly improve data quality and feature relevancy after preprocessing, eventually achieving excellent classification performance with the right combination of classification algorithms. [Copyright &y& Elsevier]

Abstract: This paper proposes a novel robust digital color image watermarking algorithm which combines color image feature point extraction, shape image normalization and QPCA (quaternion principal component algorithm) based watermarking embedding (QWEMS) and extraction (QWEXS) schemes. The feature point extraction method called Mexican Hat wavelet scale interaction is used to select the points which can survive various attacks and also be used as reference points for both watermarking embedding and extraction. The normalization shape image of the local quadrangle image of which the four corners are feature points of the original image is invariant to translation, rotation, scaling and skew, by which we can obtain the relationship between the feature images of the original image and the watermarked image which has suffered with geometrical attacks. The proposed QWEMS and QWEXS schemes which denote the color pixel as a pure quaternion and the feature image as a quaternion matrix can improve the robustness and the imperceptibility of the embedding watermarking. To simplify the eigen-decomposition procedure of the quaternion matrix, we develop a calculation approach with which the eigen-values and the corresponding eigen-vectors of the quaternion matrix can be computed. A binary watermark image is embedded in the principal component coefficients of the feature image. Simulation results demonstrate that the proposed algorithm can survive a variety of geometry attacks, i.e. translation, rotation, scaling and skew, and can also resist the attacks of many signal processing procedures, for example, moderate JPEG compression, salt and pepper noise, Gaussian filtering, median filtering, and so on. [Copyright &y& Elsevier]

Abstract: Web usage mining is the process of extracting useful information from users history databases associated to an e-commerce website. The extraction is usually performed by data mining techniques applied on server log data or data obtained from specific tools such as Google Analytics. This paper presents the methodology used in an e-commerce website of extra virgin olive oil sale called www.OrOliveSur.com. We will describe the set of phases carried out including data collection, data preprocessing, extraction and analysis of knowledge. The knowledge is extracted using unsupervised and supervised data mining algorithms through descriptive tasks such as clustering, association and subgroup discovery; applying classical and recent approaches. The results obtained will be discussed especially for the interests of the designer team of the website, providing some guidelines for improving its usability and user satisfaction. [Copyright &y& Elsevier]

Abstract: Image processing algorithms can be used in computer-aided diagnosis systems to extract features directly from digitized mammograms. Typically, two classes of features are extracted from mammograms with these algorithms, namely morphological and non-morphological features. Image texture analysis is an important technique that represents gray level properties of images used to describe non-morphological features. This technique has shown to be a promising technique in analyzing mammographic lesions caused by masses. In this paper, we evaluate texture classification using features derived from co-occurrence matrices, wavelet and ridgelet transforms of mammographic images. In particular, we propose a false positive reduction in computer-aided detection of masses. The data set consisted of 120 cranio-caudal mammograms, half containing a mass, rated as abnormal images, and half with no lesions. The following texture descriptors were then calculated to analyze the regions of interest (ROIs) texture patterns: entropy, energy, sum average, sum variance, and cluster tendency. To select the best set of features for each method, we applied a genetic algorithm (GA). In the ROIs classification stage, we used the Random Forest algorithm, a data mining technique that separates the data into non-overlapping segments. Experimental results showed that the best classification rates were obtained with the wavelet-based feature extraction using GA for selection of the most relevant features, giving an AUC=0.90. [Copyright &y& Elsevier]

Abstract: In pattern recognition, feature selection is a very important task for supervised classification. The problem consists in, given a dataset where each object is described by a set of features, finding a subset of the original features such that a classifier that runs on data containing only these features would reach high classification accuracy. A useful way to find this subset of the original features is through testor theory. A testor is defined as a subset of the original features that allows differentiating objects from different classes. Testors are very useful particularly when object descriptions contain both numeric and non-numeric features. Computing testors for feature selection is a very complex problem due to exponential complexity, with respect to the number of features, of algorithms based on testor theory. Hardware implementation of testor computing algorithms helps to improve their performance taking advantage of parallel processing for verifying if a feature subset is a testor in a single clock cycle. This paper introduces an efficient hardware–software platform for computing irreducible testors for feature selection in pattern recognition. Results of implementing the proposed platform using a FPGA-based prototyping board are presented and discussed. [Copyright &y& Elsevier]

Abstract: Local feature analysis of visual content, namely using Scale Invariant Feature Transform (SIFT) descriptors, have been deployed in the ‘bag-of-visual words’ model (BVW) as an effective method to represent visual content information and to enhance its classification and retrieval. The key contributions of this paper are first, a novel approach for visual words construction which takes physically spatial information, angle, and scale of keypoints into account in order to preserve semantic information of objects in visual content and to enhance the traditional bag-of-visual words, is presented. Second, a method to identify and eliminate similar key points, to form semantic visual words of high quality and to strengthen the discrimination power for visual content classification, is given. Third, an approach to discover a set of semantically similar visual words and to form visual phrases representing visual content more distinctively and leading to narrowing the semantic gap is specified. [Copyright &y& Elsevier]

Abstract: A novel segment confidence-based binary segmentation (SCBS) for cursive handwritten words is presented in this paper. SCBS is a character segmentation strategy for off-line cursive handwriting recognition. Unlike the approaches in the literature, SCBS is an unordered segmentation approach. SCBS is repetition of binary segmentation and fusion of segment confidence. Each repetition generates only one final segmentation point. The binary segmentation module is a contour tracing algorithm to find a segmentation path to divide a segment into two segments. A set of segments before binary segmentation is called pre-segments, and a set of segments after binary segmentation is called post-segments. SCBS uses over-segmentation technique to generate suspicious segmentation points on pre-segments. On each suspicious segmentation point, binary segmentation is performed and the highest fusion value is recorded. If the highest fusion value is greater than the one of pre-segments, the suspicious segmentation point becomes the final segmentation point for the iteration. If not, no more segmentation is required. Segment confidence is obtained by fusing mean character, lexical and shape confidences. The proposed approach has been evaluated on local and benchmark (CEDAR) databases. [Copyright &y& Elsevier]

Abstract: Extracting classification rules from data is an important task of data mining and gaining considerable more attention in recent years. In this paper, a new meta-heuristic algorithm which is called as TACO-miner is proposed for rule extraction from artificial neural networks (ANN). The proposed rule extraction algorithm actually works on the trained ANNs in order to discover the hidden knowledge which is available in the form of connection weights within ANN structure. The proposed algorithm is mainly based on a meta-heuristic which is known as touring ant colony optimization (TACO) and consists of two-step hierarchical structure. The proposed algorithm is experimentally evaluated on six binary and n-ary classification benchmark data sets. Results of the comparative study show that TACO-miner is able to discover accurate and concise classification rules. [Copyright &y& Elsevier]

Abstract: In this paper, a new intelligent method for the fault diagnosis of the rotating machinery is proposed based on wavelet packet analysis (WPA) and hybrid support machine (hybrid SVM). In fault diagnosis for mechanical systems, information about stability and mutability can be further acquired through WPA from original signal. The faulty vibration signals obtained from a rotating machinery are decomposed by WPA via Dmeyer wavelet. A new multi-class fault diagnosis algorithm based on 1-v-r SVM approach is proposed and applied to rotating machinery. The extracted features are applied to hybrid SVM for estimating fault type. Compared to conventional back-propagation network (BPN), the superiority of the hybrid SVM method is shown in the success of fault diagnosis. The test results of hybrid SVM demonstrate that the applying of energy criterion to vibration signals after WPA is a very powerful and reliable method and hence estimating fault type on rotating machinery accurately and quickly. [Copyright &y& Elsevier]