Search

Copyright of Journal of Computer Engineering & Applications is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

With the continuous development of the internet and big data, modernization and informatization are rapidly being realized in the agricultural field. In this line, the volume of agricultural news is also increasing. This explosion of agricultural news has made accurate access to agricultural news difficult, and the spread of news about some agricultural technologies has slowed down, resulting in certain hindrance to the development of agriculture. To address this problem, we apply NLP to agricultural news texts to classify the agricultural news, in order to ultimately improve the efficiency of agricultural news dissemination. We propose a classification model based on ERNIE + DPCNN, ERNIE, EGC, and Word2Vec + TextCNN as sub-models for Chinese short-agriculture text classification (E3W), utilizing the GreedySoup weighting strategy and multi-model combination; specifically, E3W consists of four sub-models, the output of which is processed using the GreedySoup weighting strategy. In the E3W model, we divide the classification process into two steps: in the first step, the text is passed through the four independent sub-models to obtain an initial classification result given by each sub-model; in the second step, the model considers the relationship between the initial classification result and the sub-models, and assigns weights to this initial classification result. The final category with the highest weight is used as the output of E3W. To fully evaluate the effectiveness of the E3W model, the accuracy, precision, recall, and F1-score are used as evaluation metrics in this paper. We conduct multiple sets of comparative experiments on a self-constructed agricultural data set, comparing E3W and its sub-models, as well as performing ablation experiments. The results demonstrate that the E3W model can improve the average accuracy by 1.02%, the average precision by 1.62%, the average recall by 1.21%, and the average F1-score by 1.02%. Overall, E3W can achieve state-of-the-art performance in Chinese agricultural news classification. [ABSTRACT FROM AUTHOR]

Text classification has always been an increasingly crucial topic in natural language processing. Traditional text classification methods based on machine learning have many disadvantages such as dimension explosion, data sparsity, limited generalization ability and so on. Based on deep learning text classification, this paper presents an extensive study on the text classification models including Convolutional Neural Network-Based (CNN-Based), Recurrent Neural Network-Based (RNN-based), Attention Mechanisms-Based and so on. Many studies have proved that text classification methods based on deep learning outperform the traditional methods when processing large-scale and complex datasets. The main reasons are text classification methods based on deep learning can avoid cumbersome feature extraction process and have higher prediction accuracy for a large set of unstructured data. In this paper, we also summarize the shortcomings of traditional text classification methods and introduce the text classification process based on deep learning including text preprocessing, distributed representation of text, text classification model construction based on deep learning and performance evaluation. [ABSTRACT FROM AUTHOR]

In the era of big data, feature engineering has proved its efficiency and importance in dimensionality reduction and useful information extraction from original features. Feature engineering can be expressed as dimensionality reduction and is divided into two types of methods, namely, feature selection and feature extraction. Each method has its pros and cons. There are a lot of studies that combine these methods. The sparse autoencoder (SAE) is a representative deep feature learning method that combines feature selection with feature extraction. However, existing SAEs do not consider feature importance during training. It causes extracting irrelevant information. In this paper, we propose an interactive guiding sparse autoencoder (IGSAE) to guide the information by two interactive guiding layers and sparsity constraints. The interactive guiding layers keep the main distribution using Wasserstein distance, which is a metric of distribution difference, and it suppresses the leverage of guiding features to prevent overfitting. We perform our experiments using four datasets that have different dimensionalities and numbers of samples. The proposed IGSAE method produces a better classification performance compared to other dimensionality reduction methods. [ABSTRACT FROM AUTHOR]

The spread of social networking sites (SNSs) has led to the development of news and its spread. In terms of momentum produced content and topics, the gathered data from online services like twitter, are from a wide range of areas. Finding an abnormality pattern as well as content oriented planning in society, is the reason why event analysis can be of great importance. Event, in social networking sites is an interesting happening in real world which causes discussion on the topic about related issues by the users of SNSs like twitter, and it is released immediately after the discussion or with a delay. Event, changes the amount of textual data, in a way that it presents related topics in a specific time period. This event is identified by time and topic, and is related with entities like people and places. The computations related to event detection in real time is a big problem in this context. A model based on deep learning is presented in this paper. Firstly, according to the labeled data, classes are formed through classification. Later, in a flow manner, unlabeled data are presented to the model. The unlabeled data are divided into the present classes according to the model which they have been trained. If the data are higher than an identified threshold, they are assigned to a new class, and when the data are lower than the threshold, they are categorized as temporary event. Innovation of the proposed method is in two issues. First, the data in this model are semi-supervised; therefore, the labeled data are used in the first phase and the rest of the data are used in the second phase. In HAN classification phase, a module titled bag of sentence was produced for exact classification of sentences, and in the second phase the abstract concept of Virtual backbone was used to enhance precision of multi label clustering. Adding these two sections using the proposed method enhanced the precision of classification and purification of the data in unlabeled data. [ABSTRACT FROM AUTHOR]

Summary: Big data is data collected with huge dimensions and continuous exponential growth over time. In recent years, big data in health care has been commonly used to predict diseases. Breast cancer is one of the most common diseases and the secondary cause of death among women. Early diagnosis of breast cancer can prevent the risk of death. Few types of research have been done on breast cancer prediction on big data. However, the traditional prediction models have less efficient in terms of accuracy and error rate. The Optimized U‐Net Convolutional neural network (OU‐NetCNN) model is proposed in this paper to overcome these challenges. Hadoop is the storage system generated to store the datasets samples for big data. The data samples from two datasets, namely BreakHis and Kaggle (Breast Histopathology Images), are preserved in this storage system. The BreakHis data are considered for further processes like pre‐processing, segmentation, feature extraction, feature selection, and classification from the stored data samples. The noise is removed from the histopathological breast images in pre‐processing using the adaptive fast peer‐group filtering (AFPGF) approach. Then the morphological operations such as erosion and dilation are used to eradicate unwanted portions and the quality of the image is enhanced using the improved balance contrast enhancement method (IBCE). Next, edges of breast images are detected using an adaptive artificial ecosystem optimization (AAEO) algorithm‐based edge detection approach in the segmentation process. The features are extracted using the Spatial Gray Level Dependence Matrix (SGLDM) and the optimal features are selected by the modified selfish herd optimization (MSHO) algorithm. Finally, the selected features are fed into the proposed OU‐NetCNN model to classify the histopathology images as benign and malignant images. This hybridization minimizes the error rate, computational complexity and over fitting issues. The simulation analysis is performed in the PYTHON tool. Two datasets, namely BreakHis and Kaggle (Breast Histopathology Images), are considered. Some of the measures such as precision, sensitivity, accuracy and F‐measure are considered to evaluate the performance of the proposed model and compared with existing approaches. [ABSTRACT FROM AUTHOR]

Smart meters provide large amounts of data and the value of this data is getting increased attention because a better understanding of the characteristics of consumers helps utilities and retailers implement more effective demand response programs and more personalized services. This paper investigates how such characteristics can be inferred from fine-grained smart meter data. A deep convolutional neural network (CNN) first automatically extracts features from massive load profiles. A support vector machine then identifies the characteristics of the consumers. Comprehensive comparisons with state-of-the-art and advanced machine learning techniques are conducted. Case studies on an Irish dataset demonstrate the effectiveness of the proposed deep CNN-based method, which achieves higher accuracy in identifying the socio-demographic information about the consumers. [ABSTRACT FROM AUTHOR]

Lung tumor is a complex disease caused due to the irregular growth of lung cells. A key factor in effective treatment planning is the early detection of lung tumor. Visual similarity between benign and malignant nodules, heterogeneity and low contrast variation are the factors that make accurate cancerous lesion recognition, a very challenging task. In this paper, Optimized Deep convolutional neural network (DCNN) and Fuzzy C‐means with Equilibrium optimizer (FCM‐EO) is proposed for classification and segmentation of CT lung images. The proposed architecture is comprised of four phases such as preprocessing, feature extraction, classification and segmentation. In preprocessing, the weighted mean histogram analysis (WMHA) is utilized to enhance the quality of images and noise removal. Hybrid Dual tree‐complex wavelet transform (DT‐CWT) with Gabor filter is proposed in feature extraction to extract the features from the preprocessed images. DCNN model is designed to classify the original images into benign and malignant images. The weight of DCNN model is updated using the Enhanced black widow optimization algorithm (EBWOA). In segmentation, FCM‐EO is introduced to identify the tumor regions and remove the outliers from the malignant images. LIDC‐IDRI dataset is utilized for the experimental analysis and MATLAB is the implementation tool. The simulation analysis is performed for both the classification and segmentation processes. Accuracy, specificity, sensitivity, precision, F‐measure, FROC, DSC, MCC, and IoU are evaluated for both these processes. The experimental results showed the proposed framework is efficient for the identification of tumor from the CT lung images. [ABSTRACT FROM AUTHOR]

In tourism-dependent cities, investigating the spatiotemporal distribution and dynamics of tourist flows is crucial for better urban planning in both steady and perturbed states. In recent years, researchers have started relying more on photo-based, geotagged social data, which offer insights about tourists, popular hotspots, and mobility patterns. However, distinguishing between tourists and locals from this data is problematic since residence information is often not provided. While previous studies rely on heuristic (e.g., period of stay) and probabilistic (Shannon entropy) approaches, this paper proposes a method for classifying tourists and residents based on machine learning (ML) algorithms and considering parameters that could explain the variability between the two (e.g., weather, mobility, and photo content). This approach was applied to Flickr users' geotagged photos taken in Tokyo's 23 special wards from July 2008 to December 2019. The results show that stacked ensemble (SE) models are superior to models based on five supervised-learning algorithms, including gradient boosting machine (GBM), generalized linear model (GLM), distributed random forest (DRF), deep learning (DL), and extremely randomized trees (XRT). Temporal entropy (TEN), mobility on workdays, and frequent visits to amusement venues and crowded places influenced how users were classified. While temporal distribution showed similar monthly/hourly patterns, spatial distribution varied. The proposed approach might pave the way for scholars to carry out future tourism research on different topics and subsequently support policymakers in the decision-making process, specifically in urban settings. [ABSTRACT FROM AUTHOR]

The oddity in large information is rising step by step so that the current programming instruments faces trouble in supervision of huge information. Moreover, the pace of the irregularity information in the immense datasets is a key imperative to the exploration business. Along these lines, this paper proposes a novel method for taking care of the large information utilizing Spark structure. The proposed method experiences two stages for arranging the enormous information, which includes highlight choice and arrangement, which is acted in the underlying hubs of Spark engineering. The proposed improvement calculation is named Rider Chaotic Biography streamlining (RCBO) calculation, which is the incorporation of the Rider Optimization Algorithm (ROA) and the standard confused biogeography-based-advancement (CBBO). The proposed RCBO-profound stacked auto-encoder utilizing Spark structure successfully handles the large information for achieving powerful huge information arrangement. Here, the proposed RCBO is utilized for choosing reasonable highlights from the monstrous dataset. Besides, the profound stacked auto-encoder utilizes RCBO for preparing so as to characterize colossal datasets. In this research we focused on problem of supervision related to big information of The Cover type Data in UCI machine learning repository. The dataset describes the forest cover set data to predict the forest cover type from cartographic variables. The dataset is multivariate in nature with number of web hits 263,361. The number of instances is 581012 with 54 numbers of attributes and the task associated for the dataset is classification. The examination of the proposed RCBO-profound stacked auto-encoder-based Spark structure utilizing the UCI AI datasets uncovered that the proposed technique beat different strategies, by procuring maximal exactness of 86.71%, dice coefficient of 92.7%, affectability of 75.2% and explicitness of 95.4% separately. [ABSTRACT FROM AUTHOR]

Big data analysis assumes a significant role in Earth observation using remote sensing images, since the explosion of data images from multiple sensors is used in several fields. The traditional data analysis techniques have different limitations on storing and processing massive volumes of data. Besides, big remote sensing data analytics demand sophisticated algorithms based on specific techniques to store to process the data in real-time or in near real-time with high accuracy, efficiency, and high speed. In this paper, we present a method for storing a huge number of heterogeneous satellite images based on Hadoop distributed file system (HDFS) and Apache Spark. We also present how deep learning algorithms such as VGGNet and UNet can be beneficial to big remote sensing data processing for feature extraction and classification. The obtained results prove that our approach outperforms other methods. [ABSTRACT FROM AUTHOR]

In the information age, the analysis of human behavioral patterns is of significant interest, and extensive data collection is crucial in the context of privacy protection. From 2011 to 2023, the authors collaborated with more than 20 volunteers to construct and collect 30,791 lifelog using mobile devices; these lifelog included a sizable amount of multi-type data containing behavioral activity information, temporal location information, textual description information, and media images. In this paper, we propose a DCP-BiLSTM (Description + Cluster + Position) And Bidirectional Long-Short Term Memory (BiLSTM) personal big data behavior classification model for the lifelog dataset, which uses text description information + GPS information classified by DB-SCAN + location information, it is validated on Liulifelog to better address the behavior prediction problem of lifelog data. Unlike the traditional text classification approach, which is based on a unique lifelog dataset, BiLSTM can combine bidirectional semantic variables comprising description information and location information into the DCP-BiLSTM model. Evaluation of the DCP-BiLSTM demonstrates that can increase the performance and accuracy of behavior prediction, and the BiLSTM that incorporates description and location information to build, not only identifies daily user activity but also corrects the bias caused by manual analysis's one-sidedness and has greater prediction performance. [ABSTRACT FROM AUTHOR]

Big data plays nowadays a major role in finance, industry, medicine, and various other fields. In this survey, 50 research papers are reviewed regarding different big data classification techniques presented and/or used in the respective studies. The classification techniques are categorized into machine learning, evolutionary intelligence, fuzzy-based approaches, deep learning and so on. The research gaps and the challenges of the big data classification, faced by the existing techniques are also listed and described, which should help the researchers in enhancing the effectiveness of their future works. The research papers are analyzed for different techniques with respect to software tools, datasets used, publication year, classification techniques, and the performance metrics. It can be concluded from the here presented survey that the most frequently used big data classification methods are based on the machine learning techniques and the apparently most commonly used dataset for big data classification is the UCI repository dataset. The most frequently used performance metrics are accuracy and execution time. [ABSTRACT FROM AUTHOR]

Classification plays a crucial role in big data, especially in e-commerce operations. Deep learning (DL) research has become a new means to provide a better solution to the problem of classification. In this paper, a deep learning spiking neural network model with the objective of building and testing a more biologically plausible network is proposed. The model is simulated using NengoDL simulator along with TensorFlow in a python environment. The performance of the model is evaluated using FashionMNIST dataset owing to its complexity than the existing MNIST dataset. The results provided better performance, and advantages are viewed in terms of area efficiency due to the use of lesser number of neurons, increasingly biologically plausible network and ease of implementation in hardware like FPGA as SNN is involved. [ABSTRACT FROM AUTHOR]

Despite the rapid improvement in dental health over the last few decades, a significant portion of our population continue seek dental care every year. Estimates show that 13% of adults seek dental care for dental infection or fatigue within four years. The Social and individual burden of this disease can be reduced by its early detection. However, the symptoms of teeth infection in the early stages are not clear, hence, it would be relatively difficult to predict teeth infections based solely on human skills and experience. Big Data (BD) technologies have a great potential in transforming dental care, as they have revolutionized other industries. In addition to reducing cost, they could save millions of lives and improve patient outcomes. This paper proposes a novel integrated prediction model that extracts hidden knowledge from radiographic datasets containing a large volume of dental X-ray images and utilizes this knowledge to predict dental infections. Initially, preprocessing techniques using morphological skeleton and mean approach is applied to eliminate noise and enhance the images. Next, Multi Scale Segmented Region (MSR) approach, Watershed Approach (WA), Sobel edge Detection (SD), Histogram based Segmentation (HS), Trainable Segmentation (TS), Dual Clustering (DC), and Fuzzy C-Means clustering (FCM) are examined for image segmentation and feature extraction. Among these methods, MSR was selected for feature extraction since it outperformed other methods in terms of accuracy, specificity, precision, recall and F1-score. Then, a set of neural network classifiers are trained to identify patterns in the extracted optimized features and predict dental infections. For this purpose, we have examined Bacterial Optimized Recurrent Neural Networks (BORNN), Deep Learning Neural Networks (DANN), Genetic Optimized Neural Networks (GONN) and Adaptive Neural Networks Algorithm (ADNN). BORNN have shown maximum accuracy and Roc value (98.1% and 0.92 respectively), and minimum error values (MSE = 0.189, MAE = 0.143). The output of the proposed integrated prediction model is fed into a dental robot who proceeds with the treatment process with high accuracy and minimum delay. The proposed prediction model was implemented using a big data analytics tool called Apache SAMOA and experimental results showed its correctness and effectiveness. [ABSTRACT FROM AUTHOR]

Nowadays, big data plays a substantial part in information knowledge analysis, manipulation, and forecasting. Analyzing and extracting knowledge from such big datasets are a very challenging task due to the imbalance of data distribution, which could lead to a biased classification results and wrong decisions. The standard classifiers are not capable of handling such datasets. Hence, a new technique for dealing with such datasets is required. This paper proposes a novel classification framework for big data that consists of three developed phases. The first phase is the feature selection phase, which uses the Whale optimization algorithm (WOA) for finding the best set of features. The second phase is the preprocessing phase, which uses the SMOTE algorithm and the LSH-SMOTE algorithm for solving the class imbalance problem. Lastly, the third phase is WOA + BRNN algorithm, which is using the Whale optimization algorithm for training a deep learning approach called bidirectional recurrent neural network for the first time. Our proposed algorithm WOA-BRNN has been tested against nine highly imbalanced datasets one of them is big dataset in terms of area under curve (AUC) against four of the most common use machine learning algorithms (Naïve Bayes, AdaBoostM1, decision table, random tree), in addition to GWO-MLP (training multilayer perceptron using Gray Wolf Optimizer), then we test our algorithm over four well-known datasets against GWO-MLP and particle swarm optimization (PSO-MLP), genetic algorithm (GA-MLP), ant colony optimization (ACO-MLP), evolution strategy (ES-MLP), and population-based incremental learning (PBIL-MLP) in terms of classification accuracy. Experimental results proved that our proposed algorithm WOA + BRNN has achieved promising accuracy and high local optima avoidance, and outperformed four of the most common use machine learning algorithms, and GWO-MLP in terms of AUC. [ABSTRACT FROM AUTHOR]

Due to the rapid growth of volume, velocity, and diversity of data termed as big data, the traditional data storage systems are inadequate nowadays. There are lots of advancement in the area of storage, processing, and analysis of data. The data are mostly stored in cloud environment and distributed computing frameworks are used to access those data. The distributed data environment in the big data processing and cloud computing imposes many security and privacy challenges. Security measures applied to data storage, analytics and processing. Firewalls are not sufficient to identify content based at-tack. A more efficient algorithm, called as PRS is proposed in the paper, which uses pattern matching technology to identify the intrusion. The working of the algorithm is based on the input data evaluated by a centralized controller based on the previous attack history and attack feature data. The performance analysis of PRS algorithm is compared with the existing pattern matching algorithms such as KMP (Knuth-Morris-Pratt Algorithm) and BM (Boyer-Moore). The simulation result shows the proposed algorithm gives a better and quick attack pattern detection compared to the existing ones. After the PRS (Pattern Recognition for Security) algorithm has identified a matching pattern, the suspected pattern is fed into a deep hypersphere model, which produces a more accurate result for identifying attack patterns. [ABSTRACT FROM AUTHOR]

In severely imbalanced datasets, using traditional binary or multi-class classification typically leads to bias towards the class(es) with the much larger number of instances. Under such conditions, modeling and detecting instances of the minority class is very difficult. One-class classification (OCC) is an approach to detect abnormal data points compared to the instances of the known class and can serve to address issues related to severely imbalanced datasets, which are especially very common in big data. We present a detailed survey of OCC-related literature works published over the last decade, approximately. We group the different works into three categories: outlier detection, novelty detection, and deep learning and OCC. We closely examine and evaluate selected works on OCC such that a good cross section of approaches, methods, and application domains is represented in the survey. Commonly used techniques in OCC for outlier detection and for novelty detection, respectively, are discussed. We observed one area that has been largely omitted in OCC-related literature is its application context for big data and its inherently associated problems, such as severe class imbalance, class rarity, noisy data, feature selection, and data reduction. We feel the survey will be appreciated by researchers working in these areas of big data. [ABSTRACT FROM AUTHOR]

Due to the insidious characteristics of network intrusion behaviors, developing an efficient intrusion detection system is still a big challenge, especially in the era of big data where the number of traffic and the dimension of each traffic feature are high. Because of the shortcomings of traditional common machine learning algorithms in network intrusion detection, such as insufficient accuracy, a network intrusion detection system based on LightGBM and autoencoder (AE) is proposed. The LightGBM-AE model proposed in this paper includes three steps: data preprocessing, feature selection, and classification. The LightGBM-AE model adopts the LightGBM algorithm for feature selection, and then uses an autoencoder for training and detection. When a set of data containing network intrusion behaviors are inputted into an autoencoder, there is a large reconstruction error between the original input data and the reconstructed data obtained by the autoencoder, which provides a basis for intrusion detection. According to the reconstruction error, an appropriate threshold is set to distinguish symmetrically between normal behavior and attack behavior. The experiment is carried out on the NSL-KDD dataset and implemented using Pytorch. In addition to autoencoder, variational autoencoder (VAE) and denoising autoencoder (DAE) are also used for intrusion detection and are compared with existing machine learning algorithms such as Decision Tree, Random Forest, KNN, GBDT, and XGBoost. The evaluation is carried out through classification evaluation indexes such as accuracy, precision, recall, F1-score. The experimental results show that the method can efficiently separate the attack behavior from normal behavior according to the reconstruction error. Compared with other methods, the effectiveness and superiority of this method are verified. [ABSTRACT FROM AUTHOR]

The rapid pace of urbanization and increasing demands for urban functionalities have led to diversification and complexity in the types of urban surface elements. The conventional approach of relying solely on remote sensing imagery for urban surface element extraction faces emerging challenges. Data-driven techniques, including deep learning and machine learning, necessitate a substantial number of annotated samples as prerequisites. In response, our study proposes a knowledge-driven approach that integrates multisource data with ontology to achieve precise urban surface element extraction. Within this framework, components from the EIONET Action Group on Land Monitoring in Europe matrix serve as ontology primitives, forming a shared vocabulary. The semantics of surface elements are deconstructed using these primitives, enabling the creation of specific descriptions for various types of urban surface elements by combining these primitives. Our approach integrates multitemporal high-resolution remote sensing data, network big data, and other heterogeneous data sources. It segments high-resolution images into individual patches, and for each unit, urban surface element classification is accomplished through semantic rule-based inference. We conducted experiments in two regions with varying levels of urban scene complexity, achieving overall accuracies of 93.03% and 97.35%, respectively. Through this knowledge-driven approach, our proposed method significantly enhances the classification performance of urban surface elements in complex scenes, even in the absence of sample data, thereby presenting a novel approach to urban surface element extraction. [ABSTRACT FROM AUTHOR]

The approval of retrieving information on the Internet originates several network securities matters. Intrusion recognition is a critical study in network security to spot unauthorized admission or occurrences on protected networks. Intrusion detection has a fully-fledged reputation in the current era. Research emphasizes several datasets to upsurge system precision and lessen the false-positive proportion. This article proposes a new intrusion detection system using big data analytics and deep learning to address some of the misuse and irregularity detection limitations. The proposed method could identify any odd activities in a network to recognize malicious or unauthorized action and permit a response during a confidentiality break. The proposed system utilizes the big data analytics platform based on parallel and distributed mechanisms. The parallel and distributed platforms improve the training time along with the accuracy. The experimentation appropriately classifies the information as either normal or abnormal. The proposed system has a recognition proportion of 96.11% that pointedly expands overall recognition accuracy related to existing strategies. [ABSTRACT FROM AUTHOR]

Big data analytics is a popular research topic due to its applicability in various real time applications. The recent advent of machine learning and deep learning models can be applied to analyze big data with better performance. Since big data involves numerous features and necessitates high computational time, feature selection methodologies using metaheuristic optimization algorithms can be adopted to choose optimum set of features and thereby improves the overall classification performance. This study proposes a new sigmoid butterfly optimization method with an optimum gated recurrent unit (SBOA-OGRU) model for big data classification in Apache Spark. The SBOA-OGRU technique involves the design of SBOA based feature selection technique to choose an optimum subset of features. In addition, OGRU based classification model is employed to classify the big data into appropriate classes. Besides, the hyperparameter tuning of the GRU model takes place using Adam optimizer. Furthermore, the Apache Spark platform is applied for processing big data in an effective way. In order to ensure the betterment of the SBOA-OGRU technique, a wide range of experiments were performed and the experimental results highlighted the supremacy of the SBOAOGRU technique. [ABSTRACT FROM AUTHOR]

A general description of modern scientometry, its main tasks, and its research methods is presented. The issues of the application of conventional machine learning and deep learning algorithms as tools of artificial intelligence in the thematic classification of scientific literature are considered. The problems and limitations of the classification of literature by sections of science in the systems of indexing and citing of scientific information are outlined. The author presents a specific example of a deep learning application for by-article thematic classification based on convolutional neural networks that was designed by scientists from the United Arab Emirates and Jordan. The article emphasizes the importance of the use of deep learning applications and models for creating correct classifications of the scientific literature that correspond to the realities of the development of science and that are capable of increasing the accuracy of calculating scientometric indicators. [ABSTRACT FROM AUTHOR]

In recent years, an increasing number of cabled Fixed Underwater Observatories (FUOs) have been deployed, many of them equipped with digital cameras recording high-resolution digital image time series for a given period. The manual extraction of quantitative information from these data regarding resident species is necessary to link the image time series information to data from other sensors but requires computational support to overcome the bottleneck problem in manual analysis. As a priori knowledge about the objects of interest in the images is almost never available, computational methods are required that are not dependent on the posterior availability of a large training data set of annotated images. In this paper, we propose a new strategy for collecting and using training data for machine learning-based observatory image interpretation much more efficiently. The method combines the training efficiency of a special active learning procedure with the advantages of deep learning feature representations. The method is tested on two highly disparate data sets. In our experiments, we can show that the proposed method ALMI achieves on one data set a classification accuracy A > 90% with less than N = 258 data samples and A > 80% after N = 150 iterations, i.e., training samples, on the other data set outperforming the reference method regarding accuracy and training data required. [ABSTRACT FROM AUTHOR]

In Mobile Performance Marketing (MPM), monetary compensation only occurs when an advertisement results in a conversion (e.g., sale of a product or service). In this work, we propose an intelligent decision support system (IDSS) to automatically select mobile marketing campaigns for users. The IDSS is based on a computationally efficient mobile user conversion prediction model that assumes a novel Percentage Categorical Pruning (PCP) categorical preprocessing and an online deep multilayer perceptron (MLP) reuse model (MLPr). Using private (nonpublicly available) business MPM data provided by a marketing company, the MLPr model outperformed an offline multilayer perceptron and a logistic regression, obtaining a high quality class discrimination when applied to sampled (85% to 92%) and complete (90% to 94%) data. In addition, the MLPr compared favorably with other machine learning (ML) models (e.g., Random Forest, XGBoost), as well as with other deep neural networks (e.g., diamond shaped). Moreover, we designed two strategies (A — best campaign selection; and B — random selection among the top candidate campaigns) to build the IDSS, in which the predictive deep learning model is used to perform a real-time selection of advertisement campaigns for mobile users. Using recently collected big data (with millions of redirect events) from a worldwide MPM company, we performed a realistic IDSS evaluation that considered three criteria: response time, potential profit and advertiser diversity. Overall, competitive results were achieved by the IDSS B strategy when compared with the current marketing company ad assignment method. [ABSTRACT FROM AUTHOR]

The current digitalized world has emerged with the advances of big data, such as online learning, business, marketing, etc. So reviewing the customer opinion in each field is more critical in enabling smart services. Several neural models are implemented diversely to characterize the customer opinion from the gained review of the user or customer. But, those models have met some difficulties detecting the aspect terms and sentiment values. So, this proposed article has planned to design a novel Strawberry Deep Belief Neural Mechanism (SDBNM) to classify the aspect and sentiment values from the trained datasets. In addition, to evaluate the performance of the designed approach, three datasets are adopted: product, trip review, and Twitter data. Initially, after the data training process, the error is extracted from the database in the preprocessing model of SDBNM. Hereafter, the error-free data is imported to the dense frame of the SDBNM to specify the aspect terms and sentiment values. At last, the parameters are estimated and compared with old approaches and have earned the best results for classification rate. [ABSTRACT FROM AUTHOR]

Diabetes is a metabolic disorder in which the body is unable to properly regulate blood sugar levels. It can occur when the body does not produce enough insulin or when cells become resistant to insulin's effects. There are two main types of diabetes, Type 1 and Type 2, which have different causes and risk factors. Early detection of diabetes allows for early intervention and management of the condition. This can help prevent or delay the development of serious complications associated with diabetes. Early diagnosis also allows for individuals to make lifestyle changes to prevent the progression of the disease. Healthcare systems play a vital role in the management and treatment of diabetes. They provide access to diabetes education, regular check-ups, and necessary medications for individuals with diabetes. They also provide monitoring and management of diabetes-related complications, such as heart disease, kidney failure, and neuropathy. Through early detection, prevention and management programs, healthcare systems can help improve the quality of life and outcomes for people with diabetes. Current initiatives in healthcare systems for diabetes may fail due to lack of access to education and resources for individuals with diabetes. There may also be inadequate follow-up and monitoring for those who have been diagnosed, leading to poor management of the disease and lack of prevention of complications. Additionally, current initiatives may not be tailored to specific cultural or demographic groups, resulting in a lack of effectiveness for certain populations. In this study, we developed a diabetes prediction system using a healthcare framework. The system employs various machine learning methods, such as K-nearest neighbors, decision tree, deep learning, SVM, random forest, AdaBoost and logistic regression. The performance of the system was evaluated using the PIMA Indians Diabetes dataset and achieved a training accuracy of 82% and validation accuracy of 80%. [ABSTRACT FROM AUTHOR]

High-Performance Computing (HPC) has recently been attracting more attention in remote sensing applications due to the challenges posed by the increased amount of open data that are produced daily by Earth Observation (EO) programs. The unique parallel computing environments and programming techniques that are integrated in HPC systems are able to solve large-scale problems such as the training of classification algorithms with large amounts of Remote Sensing (RS) data. This paper shows that the training of state-of-the-art deep Convolutional Neural Networks (CNNs) can be efficiently performed in distributed fashion using parallel implementation techniques on HPC machines containing a large number of Graphics Processing Units (GPUs). The experimental results confirm that distributed training can drastically reduce the amount of time needed to perform full training, resulting in near linear scaling without loss of test accuracy. [ABSTRACT FROM AUTHOR]

Many current intrinsically interpretable machine learning models can only handle the data that are linear, low-dimensional, and relatively independent attributes and often with discrete attribute values, while the models that are capable of handling high-dimensional nonlinear data, like deep learning, have very poor interpretability. Based on the geometric characteristics, a new idea of accurately wrapping the data region with minimum-volume geometry is proposed for pattern classification. The Variance Hyper Rectangle (VHR) model presented in this paper is a realization of the idea. The VHR model uses the minimum-volume hyper rectangles, obtained through projection variance calculation, to wrap the regions occupied by a category of data, hence it has strong and clear geometric interpretability. In addition, the VHR model is well suited for large data volume, as it approaches the linear complexity in both time and space. Extensive qualitative and quantitative experiments are performed on seven real-world data sets, demonstrating that VHR outperforms the state-of-the-art interpretable methods while running quickly. [Display omitted] • VHR has strong geometric interpretability, and is much reliable and trustworthy. • VHR can provide a clear range of values in each direction for a category of data. • VHR naturally supports incremental learning without any extra processing. • VHR has great performance and stability, and is able to handle big data. [ABSTRACT FROM AUTHOR]

XAI requires artificial intelligence systems to provide explanations for their decisions and actions for review. Nevertheless, for big data systems where decisions are made frequently, it is technically impossible to have an expert monitor every decision. To solve this problem, the authors propose an explainability auditing method for image recognition whether the explanations are relevant for the decision made by a black box model, and involve an expert as needed when explanations are doubtful. The explainability auditing system classifies explanations as weak or satisfactory using a local explainability model by analyzing the image segments that impacted the decision. This version of the proposed method uses LIME to generate the local explanations as superpixels. Then a bag of image patches is extracted from the superpixels to determine their texture and evaluate the local explanations. Using a rooftop image dataset, the authors show that 95.7% of the cases to be audited can be detected by the proposed method. [ABSTRACT FROM AUTHOR]

The rapid increase in the number of judgment documents puts forward an urgent need for automated classification.However,there is a lack of method in existing studies that use judgment results as the subject of classification in the subdivision of civil cases,and therefore they cannot achieve accurate classification of judgment results in civil cases.In this paper,we apply deep learning technology in the field of classification of judgment results of civil cases,and obtain a model with better perfor-mance in this field through horizontal comparison of multiple deep learning models.This model is further optimized based on the data characteristics of the judgment document.After experiments,the Transformer model's macro precision rate,macro recall rate and macro F1 score in the judgment result classification are all higher than other models.By adjusting the data preprocessing process and adjusting the position embedding method of the Transformer model,the performance index of the model is increased by 1%～2%.

Skin lesion classification has recently attracted significant attention. Regularly, physicians take much time to analyze the skin lesions because of the high similarity between these skin lesions. An automated classification system using deep learning can assist physicians in detecting the skin lesion type and enhance the patient's health. The skin lesion classification has become a hot research area with the evolution of deep learning architecture. In this study, we propose a novel method using a new segmentation approach and wide-ShuffleNet for skin lesion classification. First, we calculate the entropy-based weighting and first-order cumulative moment (EW-FCM) of the skin image. These values are used to separate the lesion from the background. Then, we input the segmentation result into a new deep learning structure wide-ShuffleNet and determine the skin lesion type. We evaluated the proposed method on two large datasets: HAM10000 and ISIC2019. Based on our numerical results, EW-FCM and wide-ShuffleNet achieve more accuracy than state-of-the-art approaches. Additionally, the proposed method is superior lightweight and suitable with a small system like a mobile healthcare system. [ABSTRACT FROM AUTHOR]

Deep learning became more popular in recent years. It is widely used for different machine learning tasks. One such task is sentiment prediction on a text document. Fine-grained sentiment analysis is highly recommended since most of the researchers are focusing on binary sentiment classification. In this work, a new model which combines the benefits of both Convolutional Neural Network (CNN) and Random Forest (RF) Classifier is proposed for finegrained sentiment classification. The main idea of the proposed model is to achieve maximum accuracy for sentiment classification on large volume of data. The CBOW (Continuous Bag-of-Words) model is used for converting the text input into vector form. Convolutional Neural Network (CNN) is used to extract the features from the input vector. The fully connected layer in the Convolutional Neural Network is replaced by the Random Forest classifier. Then the extracted features are used for the classification process by Random Forest Classifier. A dropout strategy is applied to regularize the CNNRF model to avoid overfitting. Sentiment analysis is performed on product review data by using CNN and RF model separately. The result of the CNN model and RF model is compared with the result of the proposed CNNRF model. The experiment result shows that the combined CNNRF model gave high performance than independent CNN and RF models. [ABSTRACT FROM AUTHOR]

Nowadays, internet is mostly used communication tool worldwide. However, the major problem of the internet is to provide security. To provide internet security, many researches and papers have been suggested about information and network security. The commonly used system against network attacks is firewalls. In this study, a novel firewall data classification approach is presented. This approach uses 10 cases to obtain numerical results. The proposed approach consists of data acquisition from Firewall, feature selection and classification steps. Firstly, the Firewall data were gathered from a Firewall. Then, the redundant features are eliminated and these features are normalized using min–max normalization. The obtained final feature sets are forwarded to classifiers. In the cases defined, Long Short-Term Memory (LSTM), Bi-directional Long Short-Term Memory (Bi-LSTM) and Support Vector Machine (SVM) are utilized as classifiers. It was seen from the results, the deep learning approach are more successful than SVM classifier and the highest classification accuracy was calculated as 97.38% by using Bi-LSTM-LSTM hybrid network. The proposed method has several advantages and these are (1) the proposed method achieved high success rates using hybrid deep learning approaches (2) the training time of the proposed method is short (3) an intelligent network security monitoring method is presented using basic methods and deep learning. In addition, a useful approach has been presented to achieve high success rate at the end of the faster training process than traditional machine learning methods. Briefly, an intelligent monitoring system is proposed for network security. • Hybrid deep learning networks are presented for firewall data classification. • 10 cases are defined to comprehensively evaluate. • An artificial intelligent based internet security system is presented. [ABSTRACT FROM AUTHOR]

Big data analysis assumes a significant role in Earth observation using remote sensing images, since the explosion of data images from multiple sensors is used in several fields. The traditional data analysis techniques have different limitations on storing and processing massive volumes of data. Besides, big remote sensing data analytics demand sophisticated algorithms based on specific techniques to store to process the data in real-time or in near real-time with high accuracy, efficiency, and high speed. In this paper, we present a method for storing a huge number of heterogeneous satellite images based on Hadoop distributed file system (HDFS) and Apache Spark. We also present how deep learning algorithms such as VGGNet and UNet can be beneficial to big remote sensing data processing for feature extraction and classification. The obtained results prove that our approach outperforms other methods.

Summary Classical Datamining methods are facing various challenges in the era of Big Data. Between the need of fast knowledge extraction and the high flows of data acquired in small slots of time, these methods became shifted. The variability and the veracity of the Big Data perplex the Machine Learning process. The high volume of Big Data yields to a congested learning because the classic methods are designed for small sets of features. Deep Learning has recently emerged in the aim of handling voluminous data. The concept of the Deep induces the conversion of the features into a new abstracted representation in order to optimize an objective. Although the Deep Learning methods are experimentally promising, their parameterization is exhaustive and empirical. To tackle these problems, we utilize the causality and the uncertainty of the Bayesian Network in order to propose a new Deep Bayesian Network architecture. We provide a new learning algorithm for this multi‐layered Bayesian Network with latent variables. We evaluate the proposed architecture and learning algorithms over benchmark datasets. We used high‐dimensional data in order to simulate the Big Data challenges, which are imposed by the volume and veracity aspects. We demonstrate the effectiveness of our contribution under these constraints. [ABSTRACT FROM AUTHOR]

Due to the vast amount of available tracking sensors in recent years, high-frequency and high-volume streams of data are generated every day. The maritime domain is no different as all larger vessels are obliged to be equipped with a vessel tracking system that transmits their location periodically. Consequently, automated methodologies able to extract meaningful information from high-frequency, large volumes of vessel tracking data need to be developed. The automatic identification of vessel mobility patterns from such data in real time is of utmost importance since it can reveal abnormal or illegal vessel activities in due time. Therefore, in this work, we present a novel approach that transforms streaming vessel trajectory patterns into images and employs deep learning algorithms to accurately classify vessel activities in near real time tackling the Big Data challenges of volume and velocity. Two real-world data sets collected from terrestrial, vessel-tracking receivers were used to evaluate the proposed methodology in terms of both classification and streaming execution performance. Experimental results demonstrated that the vessel activity classification performance can reach an accuracy of over 96 % while achieving sub-second latencies in streaming execution performance. [ABSTRACT FROM AUTHOR]

Predicting the popularity of posts on social networks has taken on significant importance in recent years, and several social media management tools now offer solutions to improve and optimize the quality of published content and to enhance the attractiveness of companies and organizations. Scientific research has recently moved in this direction, with the aim of exploiting advanced techniques such as machine learning, deep learning, natural language processing, etc., to support such tools. In light of the above, in this work we aim to address the challenge of predicting the popularity of a future post on Instagram, by defining the problem as a classification task and by proposing an original approach based on Gradient Boosting and feature engineering, which led us to promising experimental results. The proposed approach exploits big data technologies for scalability and efficiency, and it is general enough to be applied to other social media as well. [ABSTRACT FROM AUTHOR]

Measuring exact obesity rates is challenging because the existing measures, such as body mass index (BMI) and waist-to-height ratio (WHtR), do not account for various body metrics and types. Therefore, these measures are insufficient for use as health indices. This study presents a model that accurately classifies abdominal obesity, or muscular obesity, which cannot be diagnosed with BMI. Using the model, a web-based calculator was created, which provides information on obesity by predicting healthy ranges, and obesity, underweight, and overweight values. For this study, musculoskeletal mass and body composition mass data were obtained from Size Korea. The groups were divided into four groups, and six body circumference values were used to classify the obesity levels. Of the four learning models, the random forest model was used and had the highest accuracy (99%). This enabled us to build a web-based tool that can be accessed from anywhere and can measure obesity information in real-time. Therefore, users can quickly receive and update their own obesity information without using existing high-cost equipment (e.g., an Inbody machine or a body-composition analyzer), thereby making self-diagnosis convenient. With this model, it was easy to recognize and manage health conditions by quickly receiving and updating information on obesity without using traditional, expensive equipment, and by providing accurate information on obesity, according to body types, rather than information such as BMI, which are identified based on specific body characteristics. [ABSTRACT FROM AUTHOR]