Search

GF-3 satellite is China's first C-band multi-polarized synthetic aperture radar (SAR) satellite with the 1-meter resolution, which has been widely used in various fields. Road detection for GF-3 SAR images is an important part of the application of GF-3, especially in fields of map update, target recognition, and image matching. However, speckle appears in GF-3 SAR images due to coherent imaging system and it hinders the interpretation of images seriously. Especially the detection of weak roads under strong speckle background becomes extremely difficult. As a representative of multiscale geometric analysis (MGA) tool, shearlet has the optimal sparse representation feature and strong directional orientation, which can effectively capture edge and other anisotropic feature information, and can accurately describe the sparse characteristics of GF-3 SAR images. Based on shearlet, a method for detecting weak roads under strong speckle interference is proposed. Firstly, the Frost filter is used for despeckling. Secondly, shearlet is used for road detection. Finally, morphological operations are adopted to obtain the final result. Road detection experiments on various types of GF-3 SAR images demonstrate that, the proposed method can effectively overcome the interference of speckle, and completely and smoothly detect road information, which is very suitable for the detection of weak roads under strong speckle interference of GF-3 SAR images.

We present a method for the removal of movement artifacts from the recordings of electroencephalography (EEG) signals in the context of sports health. We use a smart wearable Internet of Things-based signal recording system to record physiological human signals [EEG, electrocardiography (ECG)] in real time. Then, the movement artifacts are removed using ECG as a reference signal and the baseline estimation and denoising with sparsity (BEADS) filter algorithm for trend removal. The parameters (cut-off frequency) of the BEADS filter are optimized with respect to the number of QRS complexes detected in the reference ECG signal. Next, surrogate movement signals are generated using a linear combination of intrinsic mode functions derived from the sample movement signals by the application of empirical mode decomposition. Surrogate signals are used to test the efficiency of the BEADS method for filtering the movement-contaminated EEG signals. We provide an analysis of the efficiency of the method, extracted movement artifacts and detrended EEG signals.

The steel industry is one of the pillar industries in China. The physical and mental health of steel workers is related to the development of China's steel industry. Steel workers have long been working in shifts, high temperatures, noise, highly stressed, and first-line environments. These occupational related factors have an impact on the health of steel workers. At present, the existing hypertension risk scoring models do not include occupational related factors, so they are not applicable to the risk score of hypertension in steel workers. It is necessary to establish a risk scoring model for hypertension in steel workers. In this study, the learning vector quantization (LVQ) neural network algorithm and the FisherSVM coupling algorithm are applied to estimate the hypertension risk of steel workers, and the microscopic laws of the "tailing" phenomenon of the two algorithms are analyzed by means of graphics analysis, which can describe the influence trend of sample size change in different intervals on the classification effect. The results show that the classification accuracy of the algorithm depends on the size of the sample space. When the sample size n ≤ 30 * (k + 1), the Fisher-SVM coupling intelligent algorithm is more applicable. Because its average accuracy rate is 90.00%, the average accuracy of the LVQ algorithm is only 63.34%. When the sample size is n > 30 * (k + 1), the LVQ algorithm is more applicable. Because its average accuracy rate is 93.33%, the average accuracy of the Fisher-SVM coupling intelligent algorithm is only 76.67%. The sample size of this paper is 4422, and the prediction of LVQ neural network model is more accurate. Therefore, based on the relative importance of each risk factor obtained by this model and to establish a steel worker hypertension risk rating scale, the score greater than 18 is considered as the high risk, 12-18 is considered as the medium risk, and less than 12 is considered as the low risk. Through the example's verification, the accuracy rate of the scale is 90.50% and the effect is very good. It shows that the established scoring system can effectively assess the risk of hypertension in steel workers and provide an effective basis for primary prevention of hypertension in steel workers.

Image matching is an important problem in computer vision and many technologies based on local descriptors have been developed. In this paper, we propose a novel local features descriptor based on an adaptive neighborhood and embedding Zernike moments. Instead of a fixed-size neighborhood, a size changeable neighborhood is introduced to detect the key-points and describe the features in the frame of Gaussian scale space. The radius is determined by the scale parameter of the key-point and the dominant direction is computed based on skew distribution fitting instead of the traditional eight-direction statistics. Then a 72-dimensional features vector based on a $3\times 3$ grid is presented. A 19-dimensional vector consists of Zernike moments is applied to achieve better rotation invariance and finally contributes to a 91-dimensional descriptor. The accuracy and efficiency of proposed descriptor for image matching are verified by several numerical experiments.

With the development of e-commerce, a large amount of personalized information is produced daily. To utilize diverse personalized information to improve recommendation accuracy, we propose a hybrid recommendation model based on users' ratings, reviews, and social data. Our model consists of six steps, review transformation, feature generation, community detection, model training, feature blending, and prediction and evaluation. Three groups of experiments are performed in this paper. Experiments A are used to identify the regression algorithm used in our model, Experiments B are used to identify the model to analyze review texts and the algorithm to detect social communities, and Experiments C compare our hybrid recommendation model with conventional recommendation models, such as probabilistic matrix factorization, UserKNN, ItemKNN, and social network-based models, such as socialMF and TrustSVD. The experiment results show that recommendation accuracy can be improved significantly with our hybrid model based on review texts and social communities.

Object orientation has gained a wide adoption in the software development community. To this end, different metrics that can be utilized in measuring and improving the quality of object-oriented (OO) software have been proposed, by providing insight into the maintainability and reliability of the system. Some of these software metrics are based on cognitive weight and are referred to as cognitive complexity metrics. It is our objective in this paper to present a suite of cognitive complexity metrics that can be used to evaluate OO software projects. The present suite of metrics includes method complexity, message complexity, attribute complexity, weighted class complexity, and code complexity. The metrics suite was evaluated theoretically using measurement theory and Weyuker’s properties, practically using Kaner’s framework and empirically using thirty projects.

Rapid detection of potential threads can speed up medical examination and help to start the treatment without delays. Automatic analysis of x-ray screening is a complex, multi-step process that can be beneficial for more efficient examinations in pulmonary clinics. Traditional methods use image segmentation to cut out interesting areas for further analysis of deviations from the norm (i.e., unhealthy tissues detection). However the area is extracted as a whole part, but for sensitive and more patient oriented examinations we need approach that will be extending segmentation only with necessary elements. In this article we present research results on application of heuristic method for detection over aggregated x-ray image that comes from implemented segmentation.

We address the Lithuanian author profiling task in two dimensions (AGE and GENDER) using two deep learning methods (i.e., Long Short-Term Memory - LSTM) and Convolutional Neural Network - CNN) applied on the top of Lithuanian neural word embeddings. We also investigate an impact of the training dataset size on the author profiling accuracy. The best results are achieved with the largest datasets, containing 5,000 instances in each class. Besides, LSTM was more effective on the smaller datasets, and CNN - on the larger ones. We compare the deep learning methods with the traditional machine learning methods (in particular, Naive Bayes Multinomial and Support Vector Machine), and frequencies of elements as the feature representation). The comparison revealed that the deep learning is not the best solution for our author profiling task.

We describe the development and evaluation of the educational musical computer game for preschool children that offers a first experience with music education. The game is used as a case study to demonstrate, evaluate and discuss the principles of game design for digital game based learning. We describe the game's scenario along with the explanation of the logics and mechanics of this game, and present its evaluation from the educational point of view. The game fills the niche of the Lithuanian educational musical games, while the study results show that preschool children show great interest in exploration and creation of musical sounds thus ensuring the enrichment of educational game-based process with elements of creativity and emotional learning.

We describe a novel method based on Random Projections for construction of image saliency maps. The method identifies outliers in the 2D projections of image point features as salient image points using Random Projections and kernel density estimation. We compare the method with other known methods in the area and validated on a number of benchmark images. The robustness of the method when Gaussian blurring is applied to an image is demonstrated and evaluated using F-statistics of several image quality metrics. Application of the proposed method for image processing is discussed.

We analyze and evaluate the impact of the brand-related content published in social media from the consumer perspective, and specifically focus on the gender related differences in online brand communication. We propose a research model that describes influencing consumer engagement factors such as hedonistic and interaction with brand seeking motives; the engagement levels such as consuming and participation; and brand equity elements such as awareness/associations and loyalty. To identify the engagement of consumers into brand-related content on brand equity, we used a standardized online survey in different social media channels in which 274 respondents have participated. We confirmed statistically (using paired t-test and one-way ANOVA) that women are more influenced by the interaction with brand seeking motives and their actions in social media makes bigger impact on brand equity dimension awareness; while men were influenced more by hedonistic motives and these motives make influence on men engagement into participation engagement level. Both genders were equally engaged into consuming engagement level and the engagement makes impact on such brand equity dimension as loyalty. Significant gender differences have been observed in the influence of engagement level such as interaction with brand equity seeking motives, engagement level such as consuming and brand equity dimension such as awareness/ associations and loyalty. The gender difference results in the influence of hedonistic motives and participation into engagement level were inconclusive.

Engineering processes are of paramount importance for the industry and economy. In these we need efficient simulation and positioning models that will efficiently support human operators. One of them is a radiation heat transfer problem in an electric furnace. In this article we present simulation and positioning of its operation parameters by the use of heuristic approach, which we called modified ant lion optimizer. Results of measurements and conclusions from our research show that proposed methodology can efficiently position the system.

This paper reports comparative authorship attribution results obtained on the Internet comments of the morphologically complex Lithuanian language. We have explored the impact of machine learning and similarity-based approaches on the different author set sizes (containing 10, 100, and 1,000 candidate authors), feature types (lexical, morphological, and character), and feature selection techniques (feature ranking, random selection). The authorship attribution task was complicated due to the used Lithuanian language characteristics, nonnormative texts, an extreme shortness of these texts, and a large number of candidate authors. The best results were achieved with the machine learning approaches. On the larger author sets the entire feature set composed of word-level character tetra-grams demonstrated the best performance.

Recent scientific and theoretical studies defining brand communication in social media emphasize the significance consumer engagement in brand-related content. Popularity of brand messages and the reach of target audiences depends on consumer engagement in social media. Therefore, many business companies are seeking to increase an impact on consumers using social media analysis and consumer engagement technics. Usually, consumer actions such as likes, comments and shares in social media channels are used to estimate the popularity of brand posts. One of the factors, which has not widely analyzed before, is the impact of colors for popularity of visual brand-related posts. In this paper, we analyze the effect of colors for popularity of brand-related posts in social media. We analyze our own dataset of images collected from 35 most popular brand Facebook groups. Our results show that black, gray and brown colors were more often used in images of more popular brand-related posts.

We propose four fuzzy-logic based models of tag recommendation, which are based on the interpretation of word frequency as a fuzzy membership function, and provide experimental results of tag recommendation for a variety of text datasets using different fuzzy logic operators. The novelty of the proposed models is the use of fuzzy logic modeling concepts to define a set of tags, the use of an existing set of tags for the selection of tags to strengthen the selection of most relevant (i.e., commonly used) tags, and the possibility to use an ontology to select semantically generalized tags. A system developed using the proposed models is adaptive (adapts the recommended tags to the existing set of tags), has a feedback (after each tagging, the set of tags and the dictionary are updated), is personalized (each user develops its own set of tags), and is semantics-aware (uses an ontology to refine tags). The models are validated using five sets of texts with different topics (technology, cooking, carrier, scientific, nature) and different length.

We propose a novel noise cancellation method based on the scale-adaptive remixing and demixing of Intrinsic Mode Functions (IMFs) constructed using Empirical Mode Decomposition (EMD). The method addresses the problem of mode mixing in the EMD by performing mode demixing. An illustrative example using noisy random binary sequence is presented. The proposed approach allows achieving better denoising results than the classic first IMF discarding approach.

Ambient Assisted Living (AAL) aims to improve the quality of daily life for all humans in different periods of life. Neural-Computer Interface (NCI) can be used within AAL environments to provide alternative communication means for impaired persons bypassing the need for speech and other motor activities. By monitoring, analyzing and responding to muscular activity (EMG signals) of users, NCI systems are able to monitor, diagnose and respond to the cognitive, emotional and physical states of users in real time. In this paper we analyze and develop a speller application based on the EMG interface. We analyze requirements for developing interfaces for disabled users and interfaces of known speller applications, and describe the development of the EMG-based speller as a benchmark application. The developed speller has adaptive stimulus rate and allows word selection from dictionary. We evaluate performance and usability of the developed speller using a set of empirical (accuracy, information transfer speed, input speed), ergonomic (NASA-TLX scale) and conceptual (humanistic intelligence) attributes.

Brain -- Computer interface (BCI) systems require intensive signal processing in order to form control signals for electronic devices. The majority of BCI systems work by reading and interpreting cortically evoked electro-potentials across the scalp via an electro-encephalogram (EEG). An important factor affecting the efficiency of BCI is the number of EEG features. To reduce the number of features is an important way to improve the speed. In this paper, we consider application of discrete cosine transform (DCT) on EEG signals. DCT takes correlated input data and concentrates its energy in just first few transform coefficients. This method is used as a feature extraction step and allows data size reduction without losing important information. For classification we are using artificial neural networks with different number of hidden neurons and training functions. We conclude that the method can be successfully used for the feature extraction and dataset reduction.

Variability aspects are at the core of software product family modelling approaches. In this paper, we extend the scope of variability modelling techniques through: (a) introducing a novel concept (variation sequences), which is an extension of Jaring and Bosch's variability taxonomy; and (b) formalizing feature models. The paper presents: (1) formal definitions of basic feature modelling concepts; (2) formal description of variation sequences; (3) feature selection within variation sequences using graph colouring and modal logic notations; (4) implementing feature models using meta-programming techniques with a case study.

Intelligence of present e-Learning systems is usually static: these systems may provide learning materials of different complexity and/or in differing sequence to various learners according to their abilities, skills and learning progress, but these systems hardly themselves have the ability to evolve and learn new knowledge during their life-cycle. We consider intelligence, which is characterized by (semi-)autonomous knowledge acquisition, learning and/or reasoning in order to enable the provision of better services to the users, as a system quality attribute. We propose a framework for the extension of an existing e-Learning system with intelligence capabilities. We describe an intelligent component of an e-Learning system that is capable of enriching its local domain ontology with new concepts and relationships obtained by querying a remote knowledge base.

Modern e-learning systems are mostly implemented by computer science specialists, who often lack awareness of the pedagogical/psychological aspects of e-learning processes. However, every approach to e-learning implementation must be based on sound pedagogical background. Currently, the development of adaptive and personalised learning environments is a hot topic in the e-learning domain, nevertheless, the majority of e-learning approaches still ignores learning context as an important source of knowledge. In this paper, we focus on the development of a conceptual model of learning context for e-learning environment, which captures the technological, subject domain, pedagogical, psychological aspects of learning situation. We propose a Learning Context Model (LCM) aimed for further use while developing technological implementations of adaptive, personalised learning environments.

The success of learning objects (LO) is limited by the ability to integrate them into coherent teaching courses in order to enable the educational goal-oriented creation of competency. LO sequences must be defined that are adapted to different types of learners, their personal needs and knowledge states to allow for effective personalized learning. In this paper, we propose a method for specification of LO sequences using Sequence Feature Diagrams and describe the generation of LO sequences from Generative LOs using metaprogramming techniques.

Recognition of specific functionally-important DNA sequence fragments is considered one of the most important problems in bioinformatics. One type of such fragments are promoters, i.e., short regulatory DNA sequences located upstream of a gene. Detection of promoters in DNA sequences is important for successful gene prediction. In this paper, a machine learning method, called support vector machine (SVM), is used for classification of DNA sequences and promoter recognition. For optimal classification, 11 rules for mapping of DNA sequences into binary SVM feature space are analyzed. Classification is performed using a power series kernel function. Kernel parameters are optimized using a modification of the Nelder-Mead (downhill simplex) optimization method. The results of classification for drosophila and human sequence datasets are presented.

Recognition of specific functionally-important DNA sequence fragments is considered one of the most important problems in bioinformatics. One type of such fragments is splice-junction (intron-exon or exon-intron) sites. Detection of splice-junction sites in DNA sequences is important for successful gene prediction. In this paper, support vector machine (SVM) is used for classification of DNA sequences and splice-site recognition. For optimal classification, four position-independent k-mer frequency based methods for mapping DNA sequences into SVM feature space are analyzed. Classification is performed using SVM power series kernels. Kernel parameters are optimized using a modification of the Nelder-Mead (downhill simplex) optimization method. Precision of classification is evaluated using F-measure, which is a combination of precision and recall metrics. Best classification results are achieved using 4-mers for exon-intron dataset (78%) and 6-mers for intron-exon dataset (70%) using 4-nucleotide frequencies.

We analyse ICT-based learning and consider the problem for generation of eLearning-oriented Web pages. We propose a metamodel for analysis of the problem domain as well as a metamodel for the solution domain. We represent the problem domain as a fixed set of the pre-specified parameter values and their relationships. We describe the solution domain at a higher-level as a metaspecification, which expresses parameter-based variability and is implemented using the metaprogramming techniques. The lower-level of the model describes different aspects of the teaching content such as structure, representation and management. We deliver some details of the experimental system that implements the proposed approach.