Search

Credit card fraud is one of the biggest cybercrimes faced by users. Intelligent machine learning based fraudulent transaction detection systems are very effective in real-world scenarios. However, ...

This study analyses gross job-creation and gross job-destruction determinants at the firm level for a panel of Portuguese micro-firms across four industry sectors, using the Ordinary Leat Square and Fixed Effect econometrics model to analyse a database consisting on 15.686 micro firms, for the period going from 2010 to 2017. It was found that laggard gross job-creation, assets tangibility, financial leverage, profits, and the fact firms belong to the construction sector determine gross job-creation. Regarding gross job-destruction, it was found that this variable is determined by its laggard variable, firm’s size, worker’s tenure, and the fact the firm belongs to the hotels and restaurant sector. Finally, findings suggest that a resource-based approach explains gross job-creation and gross job-creation for micro firms by using microdata. This study contributes to the state of the art on the determinants of employment and firing at micro firms’ level as it investigates the importance of the independent variables in explaining micro firm’s labour demand in Portugal.

In the real world, the class imbalance problem is a common issue in which classifier gives more importance to the majority class whereas less importance to the minority class. In class imbalance, imbalance metrics would not be suitable to evaluate the performance of classifiers with error rate or predictive accuracy. One type of imbalance data -handling method is resampling. In this paper, three resampling methods, oversampling, under-sampling and hybrid, methods are used with different approaches for in class imbalance of two different financial data to see the impact of class imbalance ratios on performance measures of nine different classification algorithms. Aiming to achieve better change classification performance, the performance of the classification algorithms, Bayes Net, Navie Bayes, J48, Random Forest Meta-Attribute Selected Classifier, MetaClassification via Regression, Meta-Logitboost, Logistic Regression, and Decision Tree, are measured on two Canadian Banks multiclass imbalance data with the performance measures, Precision, Recall, ROC Area and Kappa Statistic, by using WEKA software. The outcome of these performance measurements compared with three different resampling methods. The results provide us with a clear picture on the overall impact of class imbalance on the classification dataset and they indicate that proposed resampling methods can also be used for in class imbalance problems

This research is a descriptive research. The purpose of this research was to analyze the relationship between POGIL model and van Hiele learning in an effort to increase the level of students' geometr ical thinking. This research was implemented in one of junior high school in Bandung. The population in this research was grade VII students. The sample consisted of 63 students. Data level of students' geometr ical thinking obtained through test techniques. Test questions are arranged based on indicators from each level. The data collection technique was carried out using the triangulation method. The data analysis technique was descriptive. S tep s in this research are identifying the problem, reviewing the literature, arranging the instrument, collecting the data, analyzing the data and making the conclussion . This research shows the relation betwenen POGIL model and van Hiele Learning. They are (1) engage dan elicit steps in POGIL discribes inquiry step in van Hiele learning, (2) explore step discribes direct orientation step, (3) explanation step discribes explication step, (4) elaborate & extend steps discribes orientation step, (4) evaluation step discribes integration step.

With the rapid change in technology, cooperative innovation based on data sharing has become an imminent tactic for enterprises to gain competitive advantages. This paper adopted a mixed method approach (case study-modelling-case study) to study firms’ co-opetition behavior based on their data analytics capabilities for innovation. We show that firms favor cooperation among peers with same capabilities, i.e., when each firm’s data level is comparable to their partners. We further establish that data transferability and incentive have high impact on cooperation decisions. Finally, we explain the evolution path of firms’ cooperation decisions and discuss the best options for them to sustain long-term growth and competitiveness. The results provide a basis for firms to decide how best to utilize big data for collaborative innovation, so as to improve customers’ product adoption and reduce costs.

Cyber security is used to protect and safeguard computers and various networks from ill-intended digital threats and attacks. It is getting more difficult in the information age due to the explosion of data and technology. There is a drastic rise in the new types of attacks where the conventional signature-based systems cannot keep up with these attacks. Machine learning seems to be a solution to solve many problems, including problems in cyber security. It is proven to be a very useful tool in the evolution of malware detection systems. However, the security of AI-based malware detection models is fragile. With advancements in machine learning, attackers have found a way to work around such detection systems using an adversarial attack technique. Such attacks are targeted at the data level, at classifier models, and during the testing phase. These attacks tend to cause the classifier to misclassify the given input, which can be very harmful in real-time AI-based malware detection. This paper proposes a framework for generating the adversarial malware images and retraining the classification models to improve malware detection robustness. Different classification models were implemented for malware detection, and attacks were established using adversarial images to analyze the model’s behavior. The robustness of the models was improved by means of adversarial training, and better attack resistance is observed.

Nowadays, almost all performed activities are saved into databases. Data mining methods such as classifiers utilize these datasets for discovering hidden patterns and rules. Proposed methods for classification problems are generally developed considering approximately balanced datasets. However, imbalanced datasets that have the unequal instance numbers in its classes emerges as a common problem in most real domains. Many approaches at the data level are proposed to enable better classification of imbalanced datasets. Differential Evolution Based Oversampling Approach For Highly Imbalanced Datasets (DEBOHID) is one of the proposed method in order to handle this issue on imbalanced datasets. DEBOHID approach utilizes the crossover and mutation processes of DE for generating new synthetic samples. The parameters used by the crossover and mutation processes affect the solution quality. Therefore, in this study, solution quality in highly imbalanced datasets for different crossover and mutation parameter values of DEBOHID approach is investigated. Experimental studies are carried out by using three classifiers and two evaluation metrics for different parameter values. The obtained results are compared with well-known approaches in the literature.

Most existing classification approaches assumed underlying training data set to be evenly distributed. However, in the imbalanced classification, the training data set of one majority class could far surpass those of the minority class. This becomes a problem because it’s usually produces biased classifiers that have a higher predictive accuracy over the majority class, but poorer predictive accuracy over minority class. One popular method recently used to rectify this is the SMOTE (Synthetic Minority Over-Sampling Technique) which combines algorithms at data level. Therefore, this paper presents a novel approach for learning and imbalanced data sets, based on a combination of the SMOTE algorithm and the boosting procedure by focusing on a two-class problem. The Bidikmisi data set is imbalanced, because the distribution of majority class examples is 15 times the number of minority class examples. All models have been evaluated using stratified 5-fold cross-validation, and the performance criteria (such as Recall, F-Value and G-Mean) are examined. The results show that the SMOTE-Boosting algorithms have a better classification performance than the AdaBoost.M2 method, as the g-mean value increases 4-fold after the SMOTE method is used. We can say that SMOTE-Boosting algorithm is quite successful when taking advantage of boosting algorithms with SMOTE. When boosting affects the accuracy of the random forest by focusing on all data classes, the SMOTE algorithm alters the performance values of the random forest only in minority classes.

Background To mitigate production impact of porcine reproductive and respiratory syndrome (PRRS) virus outbreaks, it has been common to preventively vaccinate swine breeding herds using PRRS modified live virus (MLV) vaccine. However, attenuated PRRS virus (PRRSv) may result negative impact on farm productivity. The objective of this study was to measure the immediate impact of PRRS type 2 MLV vaccine on breeding herd performance under field conditions. Eight PRRS-stable farms routinely mass vaccinating females with commercial PRRS MLV vaccines were enrolled on study. Vaccination dates were collected and weekly changes in abortions, neonatal losses, pre-weaning mortality, pigs weaned per sow, and wean-to-first-service interval were assessed for up to 6 weeks after each vaccination. A 6-week period prior to each vaccination was established as baseline. Statistical process control (SPC) analysis was conducted to detect significant productivity decreases after MLV interventions, on each farm, and a mixed regression model was used, at the aggregated data level, to assess the productivity change 6 weeks after PRRS MLV vaccinations, compared to baseline. Results Out of 65 herd-MLV vaccinations, SPC analysis detected increase on abortions 4 times (6.1%), on neonatal losses 7 times (10.7%), on pre-weaning mortality 2 times (3%), on wean-to-first-service interval 2 times (3%), and no change in total pigs weaned. On aggregated data analysis, there was no significant change in abortion rate, neonatal losses, number of pigs weaned per sow, and wean-to-first-service interval. However, there was an increase of 0.26% of pre-weaning mortality 2 weeks after vaccination compared to the baseline. Conclusions Under study conditions, individual PRRS-stable sow farms had experienced transient, and numerically small changes in productivity following PRRS type 2 MLV vaccination. There was a small increase of pre-weaning mortality 2 weeks after vaccination, but no evidence of significant production impact at aggregated data analysis for abortion rate, neonatal losses, pigs weaned per sow and wean-to-first-service interval.

In real world, the distribution of dataset is not in symmetric form. It can vary from application to application and distribution of data in that application. The un-symmetric form of this distribution is called imbalanced class distribution or skewed class distribution. So, the classification of data with skewed distribution of class can lead to the poor performance of the classifier. To solve the problem of imbalanced dataset in which the instances of one class is more than the instances of other class, there are different data level approaches for handling imbalanced classes. So, in this paper we will discuss about different data level approaches and have comparative study among them.

Lake Gegas is one of the attractive tourist destinations in South Sumatra Province, Indonesia. There are quite a lot of tourists visiting Lake Gegas. This can disturb the life of the fauna there, one of which is waterbirds. In addition, it is feared that land clearing and hunting of waterbirds will reduce waterbirds and loss of habitat in Lake Gegas. The background of this research is the data related to the diversity of aerial bird species in Lake Gegas, so data collection is needed. The purpose of this study was to analyze the height level of birds in Lake Gegas. The research method used is the roaming technique. Cruising is carried out along the shores of Lake Gegas and to the other side by boat. Data collection is done by collecting techniques through collection sheets. Data level analysis using the Shannon Wiener formula. The results found 11 species of waterbirds consisting of 4 families with a contribution value of 2.21 in the "medium" category. "Habitat" by bird habitat. The conclusion of this study is that the results of research on waterbirds in Lake Gegas are included in the "medium" category.

Predicting the frequency of insurance claims has become a significant challenge due to the imbalanced datasets since the number of occurring claims is usually significantly lower than the number of non-occurring claims. As a result, classification models tend to have a limited ability to predict the occurrence of claims. So, in this paper, we'll use various data level approaches to try to solve the imbalanced data problem in the insurance industry. We developed 32 machine learning models for predicting insurance claims occurrence {(under-sampling, over-sampling, the combination of over-and under-sampling (hybrid), and SMOTE)× (three Decision tree models, three boosting models, and two bagging models) = 32}, and we compared the models' accuracies, sensitivities, and specificities to comprehend the prediction performance of the built models. The dataset contains 81628 claims, each of which is a car insurance claim. There were 5714 claims that occurred and 75914 claims that didn't occur. According to the findings, the AdaBoost classifier with oversampling and the hybrid method had the most accurate predictions, with a sensitivity of 92.94%, a specificity of 99.82%, and an accuracy of 99.4%. And with a sensitivity of 92.48%, a specificity of 99.63%, and an accuracy of 99.1%, respectively. This paper confirmed that When analyzing imbalanced data, the AdaBoost classifier, whether using oversampling or the hybrid process, could generate more accurate models than other boosting models, Decision tree models, and bagging models.

The growing interest of digital in our daily life makes Big data essential in many fields. Today, more and more companies and communities are turning to big data management to help decision-making. Understanding and better managing big data makes it possible to collect and analyze relevant information to make predictions. However, vulnerabilities exist at all scales of the big data platforms, including at the data level. Despite the tremendous efforts and resources that have been offered by big data tools and providers, big data platforms remain vulnerable to many existing forms of attacks. Therefore, new kinds of solutions should be provided to strengthen Big data security. Predictive models are offering promising solutions for additional security layers. In this paper, we summarize and discuss contributions helping to protect big data environments using Machine learning and Deep learning. We also regroup the most sensitive security aspects that should be addressed to protect valuable data. All the contributions and dimensions were addressed through a set of security use cases, namely, malware detection, intrusion, anomaly, access control, and data ingestion controls. Furthermore, we provide comparison results of different techniques to show their efficiency.

To compare the effects of preventing radiographic progression (in its 3 components) of tocilizumab (TCZ) monotherapy with those of TCZ and methotrexate (MTX) in combination therapy (TCZ + MTX), and to evaluate possible effect modifiers in this model.Randomized trials that compared TCZ monotherapy to TCZ + MTX combination therapy for differences in radiographic progression were analyzed on an individual patient data level using mixed-effects models, and data were collected from 820 subjects with either early rheumatoid arthritis (RA) or established RA. Outcomes were classified as the absence of radiographic progression after 2 years (i.e., preventing radiographic progression) as measured by total Sharp/van der Heijde score (SHS), erosion score, and joint space narrowing (JSN) score. Effect modification by baseline joint damage, disease duration, and Disease Activity Score in 28 joints (DAS28) was studied.Overall, TCZ + MTX combination therapy was more effective in preventing radiographic progression compared to TCZ monotherapy, which was measured by total SHS score. However, in patients with early RA who had more joint damage compared to those with less joint damage at baseline (relative risk [RR] 1.02 versus RR 0.91, respectively) or in patients with a lower DAS28 score compared to those with a higher DAS28 score (RR 1.04 versus RR 0.92, respectively) at baseline, this advantage disappeared. In patients with established RA, the advantage of TCZ + MTX versus TCZ alone in the prevention of radiographic progression disappeared with a longer disease duration at baseline (RR 1.04 versus 0.83). Results of erosion scores as an outcome were in line with these findings, though findings for JSN scores were less clear.Combination therapy with TCZ + MTX is more effective in preventing radiographic progression compared to TCZ monotherapy, but the effectiveness of TCZ monotherapy may approximate the effectiveness of TCZ + MTX in patients with early RA who have more joint damage and/or a lower DAS28 at baseline and in patients with established RA who have longer disease duration.

In this paper, we go around two completely different levels of program design of a biomechanical program. First, the broadest level is the data level, where we show that we can use the whole world’s data. This is covered by the System of Systems engineering. The second and most particular level is the algorithm level. Our goal is to achieve the fastest program run we can. For this, we overview the possibilities and show an example of how a parallel paradigm accelerates our program.

Recently, high profile data breach incidents have highlighted the importance of insider Intellectual Property(IP) theft research. Matching the patterns of known attack (filtering-based or rule-based) and finding the deviation from normal behavior (anomaly-based) are two typical approaches to prevent insiders from stealing sensitive information. On the one hand, filtering-based or rule-based solutions provide accurate identification of known attacks, and thus they are suitable for IP theft prevention, but they cannot handle the insiders with in-depth knowledge of the protective measures. On the other hand, anomaly-based solutions can find unknown attacks but typically have a high false-positive rate, which limits their applicability to practice. Nowadays, more and more researchers believe that the insider attack could be improved when combining known attack pattern matching with anomaly detection technologies. Therefore, in this paper, we introduce a Data-level Hybrid Framework, dubbed as Terminator, which enabling both detection and prevention. Terminator integrates a prevention module with an anomaly detection module and uses feedback to improve the module for detection or prevention. Different from previous anomaly-based methods that could only detect anomalous activities, Terminator could detect the stealing actions proactively and take real-time actions on these actions. The effectiveness of Terminator is demonstrated by its excellent performances on a collected dataset, involving detailed information in a real-world insider network and attack data simulated by impersonating the genuine users.

Proteomics become an important research area of interests in life science after the completion of the human genome project. This scientific is to study the characteristics of proteins at the large-scale data level, and then gain a holistic and comprehensive understanding of the process of disease occurrence and cell metabolism at the protein level. A key issue in proteomics is how to efficiently analyze the massive amounts of protein data produced by high-throughput technologies. Computational technologies with low-cost and short-cycle are becoming the preferred methods for solving some important problems in post-genome era, such as protein-protein interactions (PPIs). In this review, we focus on computational methods for PPIs detection and show recent advancements in this critical area from multiple aspects. First, we analyze in detail the several challenges for computational methods for predicting PPIs and summarize the available PPIs data sources. Second, we describe the state-of-the-art computational methods recently proposed on this topic. Finally, we discuss some important technologies that can promote the prediction of PPI and the development of computational proteomics.

This paper aims at investigating the effect of local government policy on the decision of secondary school enrollment in Indonesia. It combines various datasets such as Indonesia Family Life Survey based on the household and community survey, village survey (PODES), and district data level for statistical analysis. By using a probit model, this study finds that the local government policy on education, such as pupil-to-class room ratio and education expenditure per population in secondary school age, is found to have significant impact on the parents’ decision to enroll their children in the school. Further estimation on the interaction of local government policy with female child shows that only the number of public secondary school per population variable has low impact on secondary school enrollment decision.

As regards the decision-making process, there are two general principles accepted in decision-making: 1) the decision is made in terms of gain-loss; 2) a decision depends on the context; even if the decision-making rules are known, the context of the decision may be different, which means that the meaning of the concepts involved in the decision-making rules changes. The article attempts to introduce the idea that the specification of rules and constraints of concept use should be part of the level of knowledge modeling on the principle of data level separation of the logical level of the application. Specifying constraints at the place and time of manifestation of their knowledge characterizes the context and adapts the decision model. The article addresses the issue of interest and presents examples of accounting.

The prediction of aero-engine remaining useful life (RUL) is helpful for its operation and maintenance. Aiming at the challenge that most neural networks (NNs), including long short-term memory (LSTM), cannot process the input data in different update modes based on its importance degree, a novel variant of LSTM named multicellular LSTM (MCLSTM) is constructed. The level division unit is proposed to determine the importance degree of input data, and then multiple cellular units are designed to update the cell states according to the data level. Thus, MCLSTM can well mine different levels of degradation trends. Based on MCLSTM and a deep NN (DNN), a deep learning model for RUL prediction is set up, where MCLSTM and a branch of the DNN is used to extract health indicators (HIs) of aero-engine from raw data, and the other part of the DNN is applied to generate the HIs from human-made features and predict the RUL based on the concatenated HIs. The proposed RUL prediction model is successfully applied to predict the RULs of aero-engines via the Commercial Modular Aero Propulsion System Simulation datasets, and the comparative results show that it has a better comprehensive prediction performance than the commonly-used machine learning methods.

IDENTIFYING SEA SURFACE TEMPERATURE THROUGH REMOTE SENSING. Sea surface temperature is one of the physical parameters of oceanography that is useful to determine the water quality. Sea surface temperature can be observed using remote sensing technology. Remote sensing sensors must have a thermal infrared or a passive microwave to identify sea surface temperature. NASA's Ocean Color Web has provided sea surface temperature data processed by NASA's Ocean Biology Processing Group (OBPG) team. MODIS and VIIRS are remote sensing instruments used for sea surface temperature data on NASA's Ocean Color Web. The SST data provided in the form of data level 2 and level 3 are processed in daily, weekly, monthly and even yearly.

Screening of stable metabolic biomarkers objectively measuring the underlying pathophysiological changes are vital to boost early diagnosis and treatment prognosis of inherited metabolic diseases (IMDs). In this study, gas chromatography-mass spectrometry (GC–MS) was invoked to perform urinary metabolomics analysis for IMDs and a comprehensive chemometric method was presented to screen stable metabolic biomarkers effectively. For this chemometric method, a novel hybrid perturbation framework based on partial least squares discriminant analysis (PLS-DA) was adopted, in which the data level utilized bootstrap (BS) method and the function level employed the strategy of multiple informative vector fusion (MIVF), forming an algorithm BS-MIVF-PLSDA. Investigated by two common IMDs (methylmalonic acidemia and propionic acidemia), GC–MS allowed the detection of a rich profile of metabolites, characterizing subtle differences between disease and healthy control groups. The proposed chemometric method BS-MIVF-PLSDA exhibited superiorities to simpler techniques in terms of identifying candidate metabolic biomarkers more biologically correlated to the metabolic mechanisms with desirable selection stability. Moreover, the chemometric method enabled good classification performance in discriminating between disease and healthy control groups using the identified top-ranked candidate metabolic biomarkers. All the results showed that GC–MS urinary metabolomics coupled with chemometrics was feasible to offer an efficient path to achieve stable metabolic biomarkers, supporting early diagnosis and guiding treatment in IMDs clinical practice.

The paper is focused on a functional querying in graph databases. We consider labelled property graph model and mention also the graph model behind XML databases. An attention is devoted to functional modelling of graph databases both at a conceptual and data level. The notions of graph conceptual schema and graph database schema are considered. The notion of a typed attribute is used as a basic structure both on the conceptual and database level. As a formal approach to declarative graph database querying a version of typed lambda calculus is used. This approach allows to use a logic necessary for querying, arithmetic as well as aggregation function. Another advantage is the ability to deal with relations and graphs in one integrated environment.

Geographic information and building information modelling both model buildings and infrastructure, but the way in which they are modelled is usually complimentary and BIM-GIS integration is widely considered as a way forward for both domains. For one, more detailed BIM data can feed more general GIS data and GIS data can provide the context that is necessary to BIM data. While previous studies have focused on the theoretical aspects of such an integration at a schema level, in this paper we focus on explaining the geometric and topological issues we have found while trying to develop software to realise such an integration in practice and at a data level. In our preliminary results, which are presented here, we have found that many issues for such an integration remain: handling the geometric and topological problems in BIM models, dealing with bad georeferencing and figuring out the best way to convert data between IFC and CityGML are all open issues.

In this paper, a maximum-eye-tracking CDR (MET-CDR) for minimum bit error rate (BER) is presented. The proposed CDR does not require a BER counter or eye-opening monitor to find the optimal sampling phase. The biased data-level obtained from the weighted sum of UP and DN is proposed to extract the actual eye height information considering the precursor ISI. Two error samples with small time spacing detect the current eye height and the slope of the eye height so that the CDR tracks the maximum eye height where the slope becomes zero. Measured results prove that the maximum eye height phase and the minimum BER phase matches well. A prototype receiver fabricated in 28-nm CMOS process operates at 26Gb/s with an eye-opening of 25% UI and consumes 87mW while equalizing 21dB of loss at 13GHz.

In this work, a hybrid prognostic framework is put forward to predict remaining useful life (RUL) through the fusion of multiple degradation-based sensor data. This approach is tested on measurements obtained from an industrial centrifugal pump. The prognostic framework consists of determination of predication start point, sensor selection and fusion, and prognostics steps that lead to accurate RUL prediction. This approach first applies the canonical variate analysis (CVA) method to the pump for determining the prediction start time. We also present an approach for constructing a single-valued prognostic health indicator through the fusion of fault root-case variables. Moreover, the particle filter (PF) algorithm is deployed to improve the metabolism grey forecasting model (MGFM) to realize the prediction of both RUL and uncertainties.

This research aims to increase multiplication numeracy skills use arithmetic finger method towards 6 th grade students with hearing impairment of Samantha special needs school, Serang city. The purpose of this study was seen from the student’s initial skills before and after giving the application of arithmetic finger based on the criteria for using arithmetic finger, namely the increase in student’s multiplication numeracy skills. The research method is experiment with types of research Single Subject Research (SSR) with design of A-B-A. The research subject is a student of 6 th grade at Samantha special needs school, Serang city. The research data collection is through written test and observation. The research data analysis use of descriptive statistical analysis. The result showed an use arithmetic finger method can increase multiplication numeracy skills towards 6 th grade students with hearing impairment of Samantha special needs school, Serang city. This is indicated by (1) Increased sequential level of the A-B-A phase, which is 38, 75%, 70%, and 90%, (2) Changes in data stability at each phase show data that is stable (100%) and shows an ascending direction, (3) The change in the data level between phase A1/B is (+15%) and between phases B/A2 is (+10%), (4) The percentage of overlap data obtained is 0% in the A1/B phase and B/A2 phase, and (5) Students feel happy in using the arithmethic finger method.

This paper presents a 65nm CMOS 1.62-to-l0.8Gb/s video interface receiver with fully adaptive equalizers incorporating CTLE and 2-tap DFE. Sign -sign least-mean-squares (SSLMS) algorithm is used for not only the DFE but also the CTLE adaptation to reduce power consumption and extra hardware. An un-even data level is proposed for the optimum locking of the adaptation in the presence of a pre cursor. The receiver achieves BER of 10-12 at 34dB loss channel, occupies 0.174 mm2, and consumes 37.2mW at 10.8Gb/s.

In the field of data mining, imbalanced data is widespread, and ensemble learning algorithms are a classifier that is more effective in classifying imbalanced data. However, the ensemble learning algorithm itself is not optimized for imbalanced data. Therefore, an imbalanced data processing method based on data level of ensemble learning SE-gcForest is proposed. This method introduces the data level processing idea after the Multi-grained Scanning window process of the gcForest algorithm. The SMOTE algorithm and the EasyEnsemble idea deal with imbalanced data. Experiments show that this method is more effective when the data imbalanced ratio is lower and higher.

Social media analytics is one of the most important forms of data analytics in current scenario of cyber world. To leverage the use of social media analytics for branding and image building and its negative impact on misuse of such data for scamming and defamation activities, it is necessary to understand the content posted on social media and adopt comprehensive analytical techniques to predict and evaluate relevant information. This paper discusses social media analytics at data level, its challenges and issues that arise in analysis of social media data at large. It also presents a study on analysis of twitter data, where tweets are collected in real time and machine learning and lexicon-based techniques are applied to understand the sentiments and emotions behind the text that is posted on social media. The results efficiently classify the social media users opinions into positive, negative and neutral to better understand their perspectives.

In this paper, we study the problem fusion data from multi-sensors in presence of biases. We discuss both approaches, measurement data level, and track sensor level. A previous algorithm for local track fusion using a pseudo-equation with Jacobian matrix is presented for 2D sensors. In 2D case, this algorithm worked well and gave an equivalent performance and higher computational efficiency comparing with exact Kalman filter method. We extend the algorithm to 3D sensors to know how it works in this case. It is not totally straightforward when the computation of Jacobian matrix in 3D is very complex. We give the computation true Jacobian matrix for pseudo-equation using MATLAB and also give a simpler approximation Jacobian matrix. This helps to improve computational efficiency. The simulation compares the performance of the methods: Local track fusion and measurement fusion in the other cases as two sensors, four sensors, track fusion using Jacobian matrix, approximate Jacobian matrix.

In the new economic era with innovation as the first driving factor, high-tech industry, as an important industry support for innovation, its innovation efficiency will greatly affect the regional innovation level. In order to improve the innovation efficiency of the high-tech industry, the practice of accelerating the professional agglomeration of the industry has been widely adopted. What is the impact of high-tech industry specialization agglomeration on innovation level? Based on this, this paper constructs a theoretical analysis framework between industrial specialization agglomeration and innovation efficiency, and empirically analyzes it. The research results are as follows: there is an inverted U-shaped relationship between high-tech industry specialization agglomeration and innovation efficiency, that is, there is a specialized agglomeration scale, which makes industrial innovation efficiency reach the highest level. According to the empirical results, this paper provides a data level support for the proposal of industrial agglomeration policy.

Online shopping is a form of trading using electronic devices that allows consumers to buy goods or services from sellers via the internet. Other names for these activities are: e-web-shop, e-shop, e-shop, internet shop, web-shop, web-store, online shop, and virtual shop. An online store generates purchases of products or services at retailers or shopping centers, which are referred to as business-to-consumer (B2C) online shopping. n another process where a business buys from another business, it is called business-to-business (B2B). Nowadays online shopping has become more sophisticated with trading via mobile phones (m-commerce). Cellular phones have been optimized with an application to buy from online sites. In this study, we proposed a data level approach and feature selection techniques as a solution for the classification of imbalanced data. The imbalance class classification is one of the classic problems in the field of artificial intelligence, especially for classification in machine learning. Imbalanced data have been proven to reduce the performance of machine learning algorithms, where imbalance data means that the total data from each class is significantly different. The proposed method is evaluated using a dataset from the UCI repository and area under the curve (AUC) as the main evaluation. The results have shown that the proposed method produces good performance. (AUC¿ 0.8). Overall the second experiment outperformed and was better than the first and third experiments because the main evaluation in the unbalanced class classification is AUC. Therefore, it can be concluded that the proposed method produces optimal performance both for large scale data sets. Overall the second experiment outperformed and better than the first and third experiments, because the main evaluation in the unbalanced class classification was AUC.

Data augmentation is critical for deep learning-based human activity recognition (HAR) systems. However, conventional data augmentation methods, such as random-cropping, may generate bad samples that are unrelated to a particular activity (e.g. the background patches without saliency motion information). As a result, the random-cropping based data augmentation may affect negatively the overall performance of HAR systems. Humans, in turn, tend to pay more attention to motion information when recognizing activities. In this work, we attempt to enhance the motion information in HAR systems and mitigate the influence of bad samples through a Siamese architecture, termed as Motion-patch-based Siamese Convolutional Neural Network (MSCNN). The term motion patch is defined as a specific square region that includes critical motion information in the video. We propose a simple yet effective method for selecting those regions. To evaluate the proposed MSCNN, we conduct a number of experiments on the popular datasets UCF-101 and HMDB-51. The mathematical model and experimental results show that the proposed architecture is capable of enhancing the motion information and achieves comparable performance.

In1 this paper, a new parallel acceleration scheme for the High Efficiency Video Coding (HEVC) decoder based on multicore platform is proposed. In this article we will focus our research on the following two aspects: Deblocking filter module, and HEVC parallel decoding method based on task level and data level. For these two modules, we propose a series of parallel optimization algorithms. Firstly, based on previous research, we propose a more efficient parallel deblocking filtering method based on CTU units, and then, Based on a multi-core platform, we propose a HEVC parallel decoding method combining task level and data level. From the experimental data, we can see that the parallel algorithm based on multicore platform has significantly improved the parallel speedup at different QP values, and the maximum speedup has reached 14.71. The high efficiency HEVC parallel decoder designed in this thesis can complete the real-time decoding of high HD, Ultra HD streams without any encoding tools.

The authentication of GNSS signals becomes more and more important in recent years. The Navigation Message Authentication (NMA) is one approach to provide authentication of the GNSS data level. In this work, the capability and vulnerability of NMA-authenticated Galileo E1B INAV signals are evaluated. To this end, different case studies emulating spoofing attacks are investigated. The results of these tests provide a first assessment on the capability and vulnerability of NMA-authenticated Galileo signals.

The study of psychological behavior of undergraduates is one of the key problems in modern teaching and has become a hot issue. This paper introduced the commonly psychological analysis methods in undergraduate education management. Analysis the psychological changes which thanks to daily data of students’ actual life and learning. This paper puts forward the feasibility of analyzing the psychological behavior of students from the data level, and used the data of the psychological behavior of the students, to provide a reference for the teaching and management of the undergraduates.