Search

In this paper, the authors propose a new machine learning paradigm, federated machine learning. This method produces accurate predictions without revealing private data. It requires less network traffic, reduces communication costs and enables private learning from device to device. Federated machine learning helps to build models and further the models are moved to the device. Applications are particularly prevalent in healthcare, finance, retail, etc., as regulations make it difficult to share sensitive information. Note that this method creates an opportunity to build models with huge amounts of data by combining multiple databases and devices. There are many algorithms available in this area of machine learning and new ones are constantly being created. Our paper presents a comparative analysis of algorithms: FedAdam, FedYogi and FedSparse. But we need to keep in mind that FedAvg is at the core of many federated machine learning algorithms. Data testing was conducted using the Flower and Kaggle platforms with the above algorithms. Federated machine learning technology is usable in smartphones and other devices where it can create accurate predictions without revealing raw personal data. In organizations, it can reduce network load and enable private learning between devices. Federated machine learning can help develop models for the Internet of Things that adapt to changes in the system while protecting user privacy. And it is also used to develop an AI model to meet the risk requirements of leaking client's personal data. The main aspects to consider are privacy and security of the data, the choice of the client to whom the algorithm itself will be directed to process the data, communication costs as well as its quality, and the platform for model aggregation. [ABSTRACT FROM AUTHOR]

In distributed edge storage, data storage data is allocated to network edge devices to achieve low latency, high security, and flexibility. However, traditional systems for distributed edge storage only consider individual factors, such as node capacity, while overlooking the network status and the load states of the storage nodes, thereby impacting the system's read and write performance. Moreover, these systems exhibit inadequate scalability in widely adopted wireless terminal application scenarios. To overcome these challenges, this paper introduces a software-defined edge storage model and a distributed edge storage architecture grounded in software-defined networking (SDN) and the Server Message Block (SMB) protocol. A data storage node selection and distribution algorithm is formulated based on a maldistributed decision model that comprehensively considers the network and storage node load states. A system prototype is implemented in combination with 5G wireless communication technology. The experimental results demonstrate that, in comparison to conventional distributed edge storage systems, the proposed wireless distributed edge storage system exhibits significantly enhanced performance under high load conditions, demonstrating superior scalability and adaptability. This approach effectively addresses the scalability limitation, rendering it suitable for edge scenarios in mobile applications and reducing hardware deployment costs. [ABSTRACT FROM AUTHOR]

Federated learning is a distributed machine learning framework that solves data security and data island problems faced by artificial intelligence. However, federated learning frameworks are not always secure, and attackers can attack customer privacy information by analyzing parameters in the training process of federated learning models. To solve the problems of data security and availability during federated learning training, this paper proposes an Efficient Differential Privacy Federated Learning Algorithm based on early stopping mechanism (EfficientDP-FL). This method inherits the advantages of differential privacy and federated learning and improves the performance of model training while protecting the parameter information uploaded by the client during the training process. Specifically, in the federated learning framework, this article uses an adaptive DP-FL method for gradient descent training, which makes the model converge faster than traditional stochastic gradient descent. In addition, due to model convergence, noise should be reduced accordingly. This paper introduces an early stopping mechanism to improve data availability. This paper demonstrates the performance improvement of the Efficient DP-FL algorithm through simulation experiments on real MNIST and Fashion-MNIST datasets. Experimental showthat the efficient DP-FL algorithm is significantly superior to other algorithms. [ABSTRACT FROM AUTHOR]

Getting an accurate cloud cover state is a challenging task. In the past, traditional twodimensional red-to-blue band methods have been widely used for cloud detection in total sky images. By analyzing the imaging principle of cameras, green channel has been selected to replace the 2-D red-to-blue band for total sky cloud detection. The brightness distribution in a total sky image is usually non-uniform, because of forward scattering and Mie scattering of aerosols, which results in increased detection errors in the circumsolar and near-horizon regions. This paper proposes an automatic cloud detection algorithm, "green channel background subtraction adaptive threshold" (GB SAT), which incorporates channel selection, background simulation, computation of solar mask and cloud mask, subtraction, adaptive threshold, and binarization. Several experimental cases show that the GBSAT algorithm is robust for all types of test total sky images and has more complete and accurate retrievals of visual effects than those found through traditional methods. [ABSTRACT FROM AUTHOR]

In this paper, we show how the usual change of tangent altitude associated with atmospheric refraction is inseparably connected to a variation of the observed apparent intensity, for extended and pointlike sources. We demonstrate, in the regime of weak refraction angles, that atmospheric optical dilution and image deformation are strictly concomitant. The approach leads to the integration of a simple differential equation related to the observed transmittance in the absence of other absorbing molecules along the optical path. We successfully applied the proposed method to the measurements performed by two past occultation experiments: GOMOS for stellar and ORA for solar occultations. The developed algorithm (named ARID) will be applied to the imaging of solar occultations in a forthcoming pico-satellite mission. [ABSTRACT FROM AUTHOR]

Due to advent of web technologies, amount data available has grown enormously. Information retrieval from this data thus has become most important operation. Huge portion of data are available as text and poses new challenges in information retrieval and search operations. Large texts can be grouped into clusters which process the text easily. Grouping of text based on similarity of contents is called as text clustering. Topic identification refers to recognizing topic/ideas conveyed in the text. It is necessary to extract the key idea of a text if no categorization of text exists. Topics can be identified by clustering text and then extracting keywords from clusters. There are many algorithms for text clustering. Hierarchical clustering and K-means clustering are two important clustering techniques. This survey paper discusses about various clustering algorithms and application of clustering for topic identification. Also focuses on challenges and issues of text clustering and topic identification. [ABSTRACT FROM AUTHOR]

The article announces the selection of the study "Decomposition Algorithms for Stochastic Programming on a Computational Grid," written by Jeff Linderoth and Stephen Wright by the editorial board of the periodical "Computational Optimization and Applications," for the Best Paper Award 2004. The paper describes research carried out by the authors at the Argonne National Laboratory which was supported by the National Science Foundation (NSF). The research involved the development of middleware software, the discovery of new algorithms that could exploit the power of grid platforms while not being affected too seriously by its less felicitous features and the implementation of these algorithms using the resulting codes to solve touchstone problems in optimization.

In order to obtain the knowledge information of financial texts more efficiently and make the extracted information such as entity relation attribute more accurate, this paper studies the grammatical features of financial news texts and the semantic features of How Net, and puts forward the scheme of financial information extraction based on How Net. First, the phrase matching is carried out in the dictionary. Then the neural network is used for weighting, BiLSTM is used for character vector feature enhancement training, and then conditional random field (CRF) is used to complete named entity recognition, and then the relationship extraction of entity pairs from the dependency syntax is carried out to complete the research on the construction method of knowledge extraction of text in the financial field. The experimental results show that this model is superior to the other three models in entity recognition, and the overall performance is improved by about 1.2%. In relation extraction, the accuracy and recall rate of the model algorithm adopted in this paper are improved by 5% and 1.5% respectively, which shows that the improvement of the algorithm is effective. [ABSTRACT FROM AUTHOR]

Predictive Maintenance (PM) is a major part of smart manufacturing in the fourth industrial revolution. The classical fault diagnosis approach for complex systems such as an electromechanical system is not effective. Taking the advantage of the successful implementations of PM together with Deep Learning (DL) methods replaces the conventional diagnosis methods with modern diagnosis methods. This study intends to aid experts, engineers, and technicians in different electromechanical systems in comprehending how the PM used DL methods to find the multi-fault diagnosis. In this direction, this paper presents a comprehensive review of recent works of DL techniques that are applied to PM for electromechanical systems by classifying the research according to equipment, fault, parameters, and method. To perform the review, 30 papers that are published in proceedings and journals are reviewed within a time window between the years 2016 to 2022. In the context of the electromechanical system, it is observed that motors are the most equipment selected for PM. Moreover, stator winding faults are found to be less selected than bearing for diagnosis of the unhealthy status of the motor. In terms of DL methods, the study reveals that AE, LSTM, and CNN are mostly used. In addition, poorly mixed models of DL methods are also noticed. Finally, finding the optimal design variable of the DL architecture was not widely explored. [ABSTRACT FROM AUTHOR]

Electronic Travel Aid systems are expected to make impaired persons able to perform their everyday tasks such as finding an object and avoiding obstacles easier. Among ETA devices, Camera Based ETA devices are the new one and with a high potential for helping Visually Impaired Persons. With recent advances in computer science and specially computer vision, Camera Based ETA devices used several computer vision algorithms and techniques such as object recognition and stereo vision in order to help VIP to perform tasks such as reading banknotes, recognizing people and avoiding obstacles. This paper analyses and appraises a number of literatures in this area with focus on stereo vision technique. Finally, after discussing about the methods and techniques used in different literatures, it is concluded that the stereo vision is the best technique for helping VIP in their everyday navigation. [ABSTRACT FROM AUTHOR]

The combination of computer algorithms to diagnose clinical images has attracted more and more attention. This research aims to improve the efficiency of gastric cancer (GC) diagnosis, so deep learning (DL) algorithms are tentatively used to assist doctors in the diagnosis of gastric cancer. In the experiment, the collected 3591 gastroscopic images were divided into network training set and experimental verification test set. The lesion samples in the image are all marked by many endoscopists with many years of clinical experience. In order to improve the experimental effect, 5261 endoscopic images were obtained by expanding the training set. Then the obtained training set is input into the convolutional neural network (CNN) for training, and finally get the algorithm model DLU‐Net. By inputting 598 test set samples into the CNN constructed in this paper, five results such as advanced GC, early GC, precancerous lesions, normal and benign lesions can be identified and output, with a total accuracy of 94.1%. It can be concluded that the DL algorithm model constructed in this paper can effectively identify the staging characteristics of cancer as well as gastroscopic images, greatly improve efficiency, and effectively assist physicians in the diagnosis of GC under gastroscopy. [ABSTRACT FROM AUTHOR]

Detecting occluded faces is a non-trivial problem for face detection in computer vision. This challenge becomes more difficult when the occlusion covers majority of the face. Despite the high performance of current state-of-the-art face detection algorithms, the detection of occluded and covered faces is an unsolved problem and is still worthy of study. In this paper, a deep-learning-face-detection model Niqab-Face-Detector is proposed along with context-based labeling technique for detecting unconstrained veiled faces such as faces covered with niqab. An experimental test was conducted to evaluate the performances of the proposed model using the Niqab-Face dataset. The experiment showed encouraging results and improved accuracy compared with state-of-the-art face detection algorithms. [ABSTRACT FROM AUTHOR]

Advances in computer image recognition have significantly impacted many industries, including healthcare, security and autonomous systems. This paper aims to explore the potential of improving image algorithms to enhance computer image recognition. Specifically, we will focus on regression methods as a means to improve the accuracy and efficiency of identifying images. In this study, we will analyze various regression techniques and their applications in computer image recognition, as well as the resulting performance improvements through detailed examples and data analysis. This paper deals with the problems related to visual image processing in outdoor unstructured environment. Finally, the heterogeneous patterns are converted into the same pattern, and the heterogeneous patterns are extracted from the fusion features of data modes. The simulation results show that the perception ability and recognition ability of outdoor image recognition in complex environment are improved. [ABSTRACT FROM AUTHOR]

This review explores the historical and current significance of gestures as a universal form of communication with a focus on hand gestures in virtual reality applications. It highlights the evolution of gesture detection systems from the 1990s, which used computer algorithms to find patterns in static images, to the present day where advances in sensor technology, artificial intelligence, and computing power have enabled real-time gesture recognition. The paper emphasizes the role of hand gestures in virtual reality (VR), a field that creates immersive digital experiences through the Ma blending of 3D modeling, sound effects, and sensing technology. This review presents state-of-the-art hardware and software techniques used in hand gesture detection, primarily for VR applications. It discusses the challenges in hand gesture detection, classifies gestures as static and dynamic, and grades their detection difficulty. This paper also reviews the haptic devices used in VR and their advantages and challenges. It provides an overview of the process used in hand gesture acquisition, from inputs and pre-processing to pose detection, for both static and dynamic gestures. [ABSTRACT FROM AUTHOR]

Copyright of Kirkuk Journal of Science is the property of Republic of Iraq Ministry of Higher Education & Scientific Research (MOHESR) 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.)

We present a novel multi-modal adaptive feature extraction algorithm considering both time-domain and frequency-domain modalities (AFETF), coupled with a Bidirectional Long Short-Term Memory (Bi-LSTM) network based on the Grey Wolf Optimizer (GWO) for early-stage weak fault diagnosis in bearings. Singular Value Decomposition (SVD) was employed for noise reduction, while Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) was utilized for signal decomposition, facilitating further signal processing. AFETF algorithm proposed in this paper was employed to extract weak fault features. The adaptive diagnostic process was further enhanced using Bi-LSTM network optimized with GWO, ensuring objectivity in the hyperparameter optimization. The proposed method was validated for datasets containing weak faults with a 0.2 mm crack and strong faults with a 0.4 mm crack, demonstrating its effectiveness in early-stage fault detection. [ABSTRACT FROM AUTHOR]

To address the issue of the lack of specialized data filtering algorithms for dataset production, we proposed an image filtering algorithm. Using feature fusion methods to improve discrete wavelet transform algorithm (DWT) and enhance the robustness of image feature extraction, a weighted hash algorithm was proposed to hash features to reduce the complexity and computational cost of feature comparison. To minimize the time cost of image filtering as much as possible, a fast distance calculation method was also proposed to calculate the similarity of images. The experimental results showed that compared with other advanced methods, the algorithm proposed in this paper had an average accuracy improvement of 3% and a speed improvement of at least 30%. Compared with traditional manual filtering methods, while ensuring accuracy, the filtering speed of a single image is increased from 9.9s to 0.01s, which has important application value for dataset production. [ABSTRACT FROM AUTHOR]

This paper addresses the persistent challenges in speech processing within the Air Traffic Control (ATC) domain, a field where despite extensive research, issues such as handling noisy environments, accented speech, and the need for strict adherence to standard phraseology continue to undermine conventional language models. Our study employs a hybrid approach that integrates syntactic analysis with advanced machine learning classification algorithms – Logistic Regression, Lagrangian Support Vector Machine, and Naïve Bayes. By mixing and matching algorithms tailored for specific aspects of speech processing, our approach moves away from traditional reliance on a singular integrated system, illustrating through rigorous testing with the ATCOSIM dataset that such a multifaceted strategy markedly improves command recognition accuracy and adapts more effectively to the unique linguistic features of ATC speech. Results highlight the superior performance of Logistic Regression across various command recognition categories, pointing towards a promising direction for future advancements in ATC speech recognition technologies aimed at reducing human workload and increasing automation precision. This paper explores the complexities of the required analysis techniques and underscores the necessity of employing diverse algorithms in the processing pipeline to enhance overall system accuracy. [ABSTRACT FROM AUTHOR]

For the autonomous and intelligent operation of robots in unknown environments, simultaneous localization and mapping (SLAM) is essential. Since the proposal of visual odometry, the use of visual odometry in the mapping process has greatly advanced the development of pure visual SLAM techniques. However, the main challenges in current monocular odometry algorithms are the poor generalization of traditional methods and the low interpretability of deep learning-based methods. This paper presented a hybrid self-supervised visual monocular odometry framework that combined geometric principles and multi-frame temporal information. Moreover, a post-odometry optimization module was proposed. By using image synthesis techniques to insert synthetic views between the two frames undergoing pose estimation, more accurate inter-frame pose estimation was achieved. Compared to other public monocular algorithms, the proposed approach showed reduced average errors in various scene sequences, with a translation error of 2.211 % and a rotation error of 0.418 ∘ / 100 m. With the help of the proposed optimizer, the precision of the odometry algorithm was further improved, with a relative decrease of approximately 10 % intranslation error and 15 % in rotation error. [ABSTRACT FROM AUTHOR]

In this paper, we study the finite-time stability and applications of positive switched linear delayed systems under synchronous impulse control, which includes two types of random switching and average dwell time switching. By constructing a type of linear timevarying co-positive Lyapunov functional, we first propose several new finite-time stability criteria. It should be emphasized that the linear term coefficient of the linear vector of the Lyapunov functional is adjusted to the difference between the weighting vector and the given vector. Then, we apply the obtained stability criteria to the linear time-varying delayed systems with impulsive effects. At last, three examples are given to demonstrate the validity of the obtained results, which includes the specific linear programming algorithm process. [ABSTRACT FROM AUTHOR]

Predicting operational performance enables organizations to develop operational effectiveness goals considering different combinations of resources. Measuring performance is consolidated with advances in relative efficiency analysis techniques, including data envelopment analysis (DEA) and stochastic frontier analysis (SFA), albeit these methods lack predictive capability. This paper proposes an approach for performance prediction by integrating relative efficiency measurement models with machine learning algorithms. Data analyses were conducted using data provided by the energy assessment project offered to small and medium-sized manufacturing companies in the United States (n 7,548) using sales as the output, with the inputs being the number of employees, hours of operation, electricity, natural gas, cost of electricity, and cost of natural gas. Performance was estimated differently, employing parametric (SFA) and non-parametric (DEA) methods. The prediction benchmarking process occurred by adopting machine learning algorithms: regression (LM), support vector machine (SVM), K-nearest neighbor (KNN), linear discriminant analysis (LDA), random forest (RF), and decision tree (DT). The findings showed that it is possible to identify the best prediction algorithm associated with a performance model. However, the performance prediction may differ if different strategies for measuring performance or machine learning model configurations are used. In addition, SFA-LOG and SVM had the best performance for regression, and DEA-VRS/IRS excelled with random forest; the RF algorithm was the best fit across all performance approaches. The error rate depends on the algorithm and the performance model, and the number of classes must be reduced to obtain a higher success rate. [ABSTRACT FROM AUTHOR]

This article compares two optimization methods considering random variations in design parameters. One is reliability-based design optimization, which depends on the availability of the joint probability density function. A more practical alternative is robust optimization, which does not require the estimation of failure probability. It accounts for the random response of the structure through definitions of objective functions and constraints, incorporating mean values and response variances. An important element of the algorithm involves approximating unknown responses of the structures and employing efficient statistical moment estimation methods. The kriging method was used in this paper. Additionally, the article evaluates two experimental plan techniques: the classical random sampling plan and the OLH plan. [ABSTRACT FROM AUTHOR]

The ultimate goals of Augmented Reality (AR) applications are to provide a better management and ubiquitous access to information using seamless techniques in which the interactive real world is combined with an interactive computer-generated world, creating one coherent environment. The performance of the AR algorithm in terms of brightness, rotation, and scale in the real environment had been addressed as an issue due to the controlled environment. Hence, a performance evaluation of the AR algorithm in a real controlled environment is proposed in this paper. BRISK detector, FREAK descriptor, and Hamming distance matcher algorithms are implemented in a mobile AR application in order to evaluate the AR algorithm. The mobile AR application is run on the Samsung Note 9 smartphone. The image used in the evaluation is the Graffiti image from the Mikolajczyk data set. Graffiti image is fully printed in an A4 size paper and is attached to a wall with a height of 1.5m. This performance evaluation is able to evaluate the robustness of the AR algorithm in terms of brightness value from 0 Watts per square meter up to 70 Watts per square meter. The robustness of the AR algorithm in terms of scale invariance was evaluated from the distance of 5cm up to 50cm from the input image. The AR algorithm can obtain an accuracy of 83.49%, 70.89%, and 72.78% in terms of brightness changes, scale changes, and rotation changes respectively. This work introduces a more suitable performance evaluation for an AR application in a real controlled environment. [ABSTRACT FROM AUTHOR]

Blockchain technology finds widespread application across various fields due to its key features such as immutability, reduced costs, decentralization, and transparency. The security of blockchain relies on elements like hashing, digital signatures, and cryptography. However, the emergence of quantum computers and supporting algorithms poses a threat to blockchain security. These quantum algorithms pose a significant threat to both public-key cryptography and hash functions, compelling the redesign of blockchain architectures. This paper investigates the status quo of the post-quantum, quantum-safe, or quantum-resistant cryptosystems within the framework of blockchain. This study starts with a fundamental overview of both blockchain and quantum computing, examining their reciprocal influence and evolution. Subsequently, a comprehensive literature review is conducted focusing on Post-Quantum Distributed Ledger Technology (PQDLT). This research emphasizes the practical implementation of these protocols and algorithms providing extensive comparisons of characteristics and performance. This work will help to foster further research at the intersection of post-quantum cryptography and blockchain systems and give prospective directions for future PQDLT researchers and developers. [ABSTRACT FROM AUTHOR]

Information and communication technologies are spreading rapidly due to their fast proliferation in many fields. The number of Internet users has led to a spike in cyber-attack incidents. E-commerce applications, such as online banking, marketing, trading, and other online businesses, play an integral role in our lives. Network Intrusion Detection System (NIDS) is essential to protect the network from unauthorized access and against other cyber-attacks. The existing NIDS systems are based on the Backward Oracle Matching (BOM) algorithm, which minimizes the false alarm rate and causes of high packet drop ratio. This paper discussed the existing NIDS systems and different used pattern-matching techniques regarding their weaknesses and limitations. To address the existing system issues, this paper proposes an enhanced version of the BOM algorithm by using multiple pattern-matching methods for the NIDS system to improve the network performance. The proposed solution is tested in simulation with existing solutions using the Snort and NSL-KDD datasets. The experimental results indicated that the proposed solution performed better than the existing solutions and achieved a 5.17% detection rate and a 0.22% lower false alarm rate than the existing solution. [ABSTRACT FROM AUTHOR]

3D models are essential in virtual reality, game development, architecture design, engineering drawing, medicine, and more. Compared to digital images, 3D models can provide more realistic visual effects. In recent years, significant progress has been made in the field of digital image encryption, and researchers have developed new algorithms that are more secure and efficient. However, there needs to be more research on 3D model encryption. This paper proposes a new 3D model encryption algorithm, called the 1D map with sin and logistic coupling (1D-MWSLC), because existing digital image encryption algorithms cannot be directly applied to 3D models. Firstly, this paper introduce 1D-MWSLC, which has a wider range of parameters compared to traditional 1D chaotic systems. When the parameter exceeds a specific range, the chaotic phenomenon does not weaken. Additionally, 1D-MWSLC has two control parameters, which increases the cryptosystem's parameter space. Next, 1D-MWSLC generates keystreams for confusion and diffusion. In the confusion stage, this paper use random confusion, and the keystream generates an index matrix that confuses the integer and decimal parts of the 3D model simultaneously. In the diffusion stage, this paper use parallel bidirectional diffusion to simultaneously diffuse the integer parts of the three coordinates of the 3D model. Finally, this paper verify the proposed algorithm through statistical analysis, and experimental results demonstrate that the proposed 3D model encryption algorithm has robust security. [ABSTRACT FROM AUTHOR]

In this paper, we consider the issue of digital camera identification based on images. This topic matches the area of digital forensics. The problem is well known in the literature and many algorithms based on camera's fingerprint have been proposed. In this paper, we discuss the digital camera identification based on convolutional neural networks (CNN). CNNs are state-of-the-art method in computer vision and are widely utilized in many applications. Our goal is to find out whether it is possible to speed up the process of learning the networks by the images. We conduct a set of representative experiments which show that replacing the ReLU with SELU activation function and adjusting the network's hyperparamethers (e.g. learning rate) may have a significant impact on reduction time of learning. We also consider using the dropout layer. The experiments are held on representative image dataset, consisting of many images coming from modern cameras and show effectiveness of our propositions. [ABSTRACT FROM AUTHOR]

This paper investigates the multifaceted issue of algorithmic bias in artificial intelligence (AI) systems and explores its ethical and human rights implications. The study encompasses a comprehensive analysis of AI bias, its causes, and potential remedies, with a particular focus on its impact on individuals and marginalized communities. The primary objectives of this research are to examine the concept of algorithmic bias, assess its ethical and human rights implications, identify its causes and mechanisms, evaluate its societal impact, explore mitigation strategies, and examine regulatory and community-driven approaches to address this critical issue. The research employs a multidisciplinary approach, drawing from literature reviews, case studies, and ethical analyses. It synthesizes insights from academic papers, governmental reports, and industry guidelines to construct a comprehensive overview of algorithmic bias and its ramifications. This research paper underscores the urgency of addressing algorithmic bias, as it raises profound ethical and human rights concerns. It advocates for comprehensive approaches, spanning technical, ethical, regulatory, and communitydriven dimensions, to ensure that AI technologies respect the rights and dignity of individuals and communities in our increasingly AI-driven world. [ABSTRACT FROM AUTHOR]

This paper consists of a bibliometric study that covers the topic of 3D object detection from 2022 until the present day. It employs various analysis approaches that shed light on the leading authors, affiliations, and countries within this research domain alongside the main themes of interest related to it. The findings revealed that China is the leading country in this domain given the fact that it is responsible for most of the scientific literature as well as being a host for the most productive universities and authors in terms of the number of publications. China is also responsible for initiating a significant number of collaborations with various nations around the world. The most basic theme related to this field is deep learning, along with autonomous driving, point cloud, robotics, and LiDAR. The work also includes an in-depth review that underlines some of the latest frameworks that took on various challenges regarding this topic, the improvement of object detection from point clouds, and training end-to-end fusion methods using both camera and LiDAR sensors, to name a few. [ABSTRACT FROM AUTHOR]

Profit prediction is a pivotal task in financial markets, empowering investors and traders to make informed decisions. In recent years, the advent of deep learning techniques has revolutionized the field of financial forecasting, offering the potential to extract intricate patterns and relationships from vast and complex datasets. This paper presents an innovative approach to profit prediction using Long Short-Term Memory (LSTM) networks, a specialized type of Recurrent Neural Network (RNN). LSTMs excel at capturing long term dependencies in sequential data, making them well-suited for modeling the dynamics of the financial markets. The core of the paper lies in the practical application of LSTM model architecture specially tailored for profit prediction. This includes defining the input layer, LSTM layers, fully connected layers and the output layer. The training and validation process is elucidated, covering data splitting, model training, validation techniques and hyper parameter tuning to enter ensure the model performance. The paper also explores the practical application of the LSTM-based profit prediction algorithm through a case study involving real-world financial data. Evaluation metrics such as Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) are employed to assess the algorithm's predictive accuracy and effectiveness. Additionally, the paper addresses risk assessment, a critical aspect of profit prediction in financial markets. It sheds light on the promising potential of LSTM-based profit prediction algorithms as a powerful tool for financial forecasting. It summarizes key findings, acknowledges limitations and challenges, and outlines future directions for improving the algorithm, including incorporating additional data sources and fine-tuning hyper parameters. The presented approach offers a significant advancement in the realm of profit prediction, enabling investors and traders to make more informed and data-driven decisions in an ever-evolving financial landscape. [ABSTRACT FROM AUTHOR]

Solving arithmetic word problems (AWPs) that involve deep implicit relations can be quite challenging. However, the paper proposed two approaches to tackle this issue. The first approach used the modifier-to-matrix (MTM) model to extract noun modification components from the problem text. Specifically, a missing entity recovery (MER) model translated explicit expressions into a node dependency graph (NDG). The nodes on the graph then recursively acquired connections from the knowledge base through the MER model until the goal was achieved with a known quantity. The solving engine then selected the appropriate knowledge as the prompt. The second approach proposed in the paper was a comprehensive one that combined explicit and implicit knowledge to enhance reasoning abilities. The experimental results of the dataset demonstrate that the proposed algorithm is superior to the baseline algorithms in solving AWPs that require deep implicit relations. [ABSTRACT FROM AUTHOR]

The digital economy is developing rapidly worldwide, which is of great strategic significance in leading and driving the process of high-quality and high-efficiency development of the economy. It has gradually become a new engine to promote China's economic growth. Therefore, it is vital to establish a digital economy indicator system. In this context, according to the connotation of digital economy, comprehensively consider the four dimensions of digital foundation, digital application, digital innovation and digital benefit, and build a full-scale digital economy indicator system. However, some previous literature proposed operating the K-means clustering algorithm in optimizing the index system. The algorithm, susceptible to the initial selection of cluster center, generates completely different clustering results with other random seed points and is thus unsuitable for optimizing the index system. Based on this, the advantages of GRA, SA, and K-means algorithms are combined to propose a K-means-SA algorithm that can obtain the global optimal solution. Then, by combining the K-means SA algorithm and the rough set algorithm, the constructed index system is further optimized. Ultimately, it establishes a set of all-round, multi-system and multi-dimensional digital economy index systems, which is of great reference significance for formulating relevant policies, provides scientific index support for the subsequent extension of digital economy theory, and promotes the benign growth of the worldwide economy. Plain language summary: The importance of constructing and optimizing the digital economy indicator system In today's rapid development of the digital economy, a set of scientific, feasible, and practical application values of the digital economy evaluation index system is constructed to evaluate better and promote the development of the digital economy. Based on the four dimensions of digital foundation, digital application, digital innovation, and digital benefit, this paper constructs a comprehensive evaluation index system for the development level of the digital economy. In addition, by combining the advantages of GRA, SA, and K-means algorithms, a K-means SA algorithm can obtain the global optimal solution proposed. Then, the K-mean-SA algorithm is combined with the rough set algorithm to optimize the constructed index system further. Finally, a comprehensive, multisystem, and multi-dimensional evaluation index system for the development level of the digital economy is established, which provides scientific support for the development of digital economy theory and promotes the healthy development of the digital economy. Although this paper collects indicators related to the digital economy as comprehensively as possible based on existing studies, there are still some limitations in the selection of needles due to the strong integration of the digital economy. [ABSTRACT FROM AUTHOR]

This article proposes the deep learning algorithm- Convolutional Neural Networks (CNN) for both Glioma tumor classifications and diagnosis process. This proposed CNN architecture is derived from the conventional CNN architecture to obtain the optimum classification and diagnosis accuracy. This proposed CNN architecture is derived from the conventional system for obtaining the high classification and diagnosis performance. This proposed methodology stated in this paper uses BRATS 2015 open access dataset for obtaining the brain Magnetic Resonance Image (MRI) for tumor region detection. The proposed methodology stated in this paper for tumor diagnosis achieves 97.7% of Jaccard Index (J) and 83.8% of Dice Similarity Index (DSI) and 99.025 of Diagnosis Rate (DR) using CNN algorithm.. [ABSTRACT FROM AUTHOR]

Nowadays, the consumption is the most significant factor in the economy. The consumption requires different items, which must be produced by the industry. Production processes are realized mainly on workplaces, so their importance is inevitably. This paper presents an overview about the planning process of material handling machines, especially for the production workplaces. The research, shown in this paper, targets to select the most suitable material handling equipment for different production workplaces. The topic presented in this paper deals with the building of selection algorithm for the exclusion of non-applicable equipment types. To show the useability of the concept an example will also be presented. [ABSTRACT FROM AUTHOR]

Pure electric vehicle (PEV) equipped with a dual-motor coupling drive system can make full use of the high efficiency working range of the motor in order to improve vehicle efficiency. In order to further expand the application range of the system and to improve its practical application, this paper designs and proposes a new dynamic coupling drive system of three axis-double working modes, which is based on the Simpson planetary gear train. The new system adopts two planetary gears (P1 and P2), and the two sun gears of double rows, planetary carrier of P1 and gear ring of P2 are bunded. The power output of the P1 gear ring (mode 1) and P2 planetary carrier (mode 2) is realized by a controlling wet clutch. This paper adopts the linear interpolation method, least square method and 5-fold CV cross validation method to establish the full load speed characteristics and efficiency characteristics models of two motors (13 and 30 kW). This paper proposes an optimization design method based on an improved simulated annealing (I-SA) algorithm for new system parameter matching and working mode switching strategy determination. The results show that the modeling accuracy of the two motors is high, and the mean value of MAPE is 4.337%. The proposed optimization design method achieves the demand goal of PEV effectively. The I-SA algorithm has good effectiveness and fast convergence, the mean efficiency of the optimized PEV is 83.91% under all working conditions, the maximum speed is 142.56 km/h and the power utilization rate of the dual-motor is 100%. This study proposes a new hardware system and a design optimization method on software and provides a direct reference for the research of PEV drive systems by combining hardware with software. [ABSTRACT FROM AUTHOR]

An introduction is presented in which the guest editors discuss various reports within the issue on topics including one on good edit similarity learning by loss minimization, one concerning intrinsic dimensionality estimators, and one on generating feature spaces for linear algorithms consisting sparse kernel slow feature analysis.

The article focuses on a study which aims at the application of an improved cloud detection algorithm for FengYun-3A/Visible and InfRared Radiometer (FY-3A/VIRR) over China's northwest region. In the study, the Cloud and Aerosol Unbiased Decision 25 Intellectual Algorithm (CLAUDIA) is applied to FY-3A/VIRR data. It concludes increased probability of detecting clear-sky regions over snow using the Moderate Resolution Imaging Spectroradiometer (MODIS).

In recent years, the road visibility detection method based on video has been paid more and more attention. It has overcome the deficiency of laser visibility meter to some extent. Deep learning has a good effect in image processing and analysis. This paper firstly analyzes the current situation of deep learning, and then compares DenseNet and ResNet to propose a visibility estimation model based on deep DenseNet. The model firstly integrates airport video data and visibility data. Secondly, the DenseNet algorithm is used to automatically extract the features of the airport data set. Finally, Softmax classifier is constructed to evaluate the visibility accuracy. They reduce the problem of disappearing gradient, enhance feature propagation, encourage functional reuse, and greatly reduce the number of parameters, well train the deep model, has a good visibility estimation effect. On this basis, this paper based on Canny operator lane dividing line extraction edge extraction and visibility analysis based on edge detection, and do the corresponding test. Finally, a video visibility analysis model based on Kalman filter is built based on the given data, and Gaussian process regression model is used to predict the fog change trend. [ABSTRACT FROM AUTHOR]

This paper proposes a blockchain-based decentralized collaborative learning method for the Industrial Internet environment to solve the trust and security issues in Federated Learning. Deploy a decentralized network for collaborative learning based on the alliance chain, design a block data structure suitable for asynchronous learning, and model three stages of computing event triggering, computing task distribution, and computing result integration for cross-domain device collaborative learning. List the critical steps for network deployment, including inspection, tearing down old networks, creating organizational encryption material, creating channels, and deploying chaincode. It also introduces the development of crucial chaincode such as initialization, creation, query, and modification. Finally, the correlation between the number of data pieces of the network, the number of communications, and the time of communications are analyzed through experiments. This paper also proposes a decentralized asynchronous collaborative learning algorithm, develops chaincode middleware between the blockchain network and Artificial Intelligence training, and conducts experimental analysis on the industrial steam volume prediction data set in thermal power generation. The performance on the data set, and the experimental results prove that the asynchronous collaborative learning algorithm proposed in this paper can achieve a good convergence effect. It is also compared with the single-machine single-card regression prediction algorithm, proving that the proposed model has better generalization. [ABSTRACT FROM AUTHOR]

Homogeneous and isotropic thick spheres loaded with constant internal and/or external pressures can not be economically designed because the maximum equivalent stress is a local value. It was analytically demonstrated that, by neglecting the body loads, a functionally graded material (FGM) may be characterized in linear static analyses by two material constants: Young's modulus E(r) and Poisson's ratio ν(r). If these two functions are both known, the solution of the problem, displacement and stress distributions may be relatively easy obtained. For the inverse problem, in which a desired stress combination distribution is imposed, finding of E(r) and ν(r) is more difficult, if such a solution exists. More than that, if the solution exists, it is not unique, because two unknown functions are involved. For ν(r) = const., analytical solutions are available for E(r), but only for two particular stress conditions. In this paper, the inverse problem is solved iteratively using a finite element model and an algorithm of stress uniformization developed by the authors of this paper is proposed. In this original approach, the existing solutions were reproduced as a verification and afterwards new solutions were obtained for the remaining classical theories of resistance. The new obtained solutions were also verified by using the analytical solutions of the direct problem. [ABSTRACT FROM AUTHOR]

This paper considers an assembly problem. Let there be two interlocking parts, only one of which may be cut into pieces. How many pieces should we cut the divisible part into to separate the parts using a sequence of rigid-body motions? In this initial exploration, we primarily consider 2-D polygonal parts. This paper presents an algorithm that computes a lower bound on the number of pieces that the divisible part must be cut into. This paper also presents a complete algorithm that constructs a set of cuts and a motion plan for disassembly, yielding an upper bound on the required number of pieces. Applications of the future extension of this paper to 3-D may include robot self-assembly, interlocking 3-D model design, search-and-rescue, packaging, and robotic surgery. Note to Practitioners—This paper presents an opposite problem of immobilization or caging. Given two interlocking parts, only one of which may be cut into pieces. How can we unimmobilize or uncage one from another? We explore this problem in two aspects: the lower bound and the upper bound on the number of pieces that the divisible part must be cut into. This paper is an early theoretical exploration of this problem. We verified our methods in a virtual 2-D environment instead of building physical structures. Extension of this paper to 3-D could have many applications, including robot self-assembly, 3-D fabrication model design, search-and-rescue, packaging, and robotic surgery. [ABSTRACT FROM AUTHOR]

After the revival of deep learning in computer vision in 2012, SAR ship detection comes into the deep learning era too. The deep learning-based computer vision algorithms can work in an end-to-end pipeline, without the need of designing features manually, and they have amazing performance. As a result, it is also used to detect ships in SAR images. The beginning of this direction is the paper we published in 2017BIGSARDATA, in which the first dataset SSDD was used and shared with peers. Since then, lots of researchers focus their attention on this field. In this paper, we analyze the past, present, and future of the deep learning-based ship detection algorithms in SAR images. In the past section, we analyze the difference between traditional CFAR (constant false alarm rate) based and deep learning-based detectors through theory and experiment. The traditional method is unsupervised while the deep learning is strongly supervised, and their performance varies several times. In the present part, we analyze the 177 published papers about SAR ship detection. We highlight the dataset, algorithm, performance, deep learning framework, country, timeline, etc. After that, we introduce the use of single-stage, two-stage, anchor-free, train from scratch, oriented bounding box, multi-scale, and real-time detectors in detail in the 177 papers. The advantages and disadvantages of speed and accuracy are also analyzed. In the future part, we list the problem and direction of this field. We can find that, in the past five years, the AP50 has boosted from 78.8% in 2017 to 97.8 % in 2022 on SSDD. Additionally, we think that researchers should design algorithms according to the specific characteristics of SAR images. What we should do next is to bridge the gap between SAR ship detection and computer vision by merging the small datasets into a large one and formulating corresponding standards and benchmarks. We expect that this survey of 177 papers can make people better understand these algorithms and stimulate more research in this field. [ABSTRACT FROM AUTHOR]

The increasing value of facilities, on the one hand, and the complexity of the equipment used in them, on the other, have increased the importance of planning for the maintenance of facilities, especially for companies whose facilities are located in different locations. In this paper, a new hybrid model is presented to optimize facility maintenance scheduling by a combination of Genetic Algorithms (GA), Particle Swarm Optimization (PSO), and the Monte Carlo Simulation for organizing facilities located in different locations as well as determining the optimum number of crews with mechanical, electrical and simple skills. The main contributions of this paper include: (a) optimizing the number of crews by different skills in the first stage; (b) evaluation of fitness value for each solution through the Monte Carlo Simulation model; and (c) scheduling by considering different failure rates for different facilities in different locations. In order to evaluate the performance of the proposed model, the model has been compared with Golpira model, the results of which demonstrate that it is possible to reduce the cost by just over 39% and reduce Mean Time Between Failures (MTBF) by over half. [ABSTRACT FROM AUTHOR]

Food and fluid intake monitoring are essential for reducing the risk of dehydration, malnutrition, and obesity. The existing research has been preponderantly focused on dietary monitoring, while fluid intake monitoring, on the other hand, is often neglected. Food and fluid intake monitoring can be based on wearable sensors, environmental sensors, smart containers, and the collaborative use of multiple sensors. Vision-based intake monitoring methods have been widely exploited with the development of visual devices and computer vision algorithms. Vision-based methods provide non-intrusive solutions for monitoring. They have shown promising performance in food/beverage recognition and segmentation, human intake action detection and classification, and food volume/fluid amount estimation. However, occlusion, privacy, computational efficiency, and practicality pose significant challenges. This paper reviews the existing work (253 articles) on vision-based intake (food and fluid) monitoring methods to assess the size and scope of the available literature and identify the current challenges and research gaps. This paper uses tables and graphs to depict the patterns of device selection, viewing angle, tasks, algorithms, experimental settings, and performance of the existing monitoring systems. [ABSTRACT FROM AUTHOR]

In industrial practice, is sometimes needed to manufacture specific parts without blueprints. This part of engineering process is known as reverse engineering. Usually, this assumes that an existing piece is measured or scanned and based on this result is made a new part. If the part can be machined with universal cutting tools the manufacturing process don't rise problems, but in case of parts which needs specific tools, these must be designed to be suitable for machining the piece. In this paper, a tool profiling algorithm is proposed, which is based on the capacity to measure and represent some surfaces in a methodical cloud of points form. The proposed profiling algorithm is based on the "virtual pole" method. This method is used together with the intermediate surface theorem for profiling tools characterized by a helical primary peripheral surface. This article will represent a continuation of a research which was been published in 2021, in Bulletin of Polytechnic Institute from Jassy. In that published article, the profiling of the hob mill tool was presented in a theoretical way and the "virtual pole" theorem was applied for an any type of profiles. The proposed algorithm allowed the determination of the tool's profile reciprocally enveloping with a piece's profile known in discrete form. Now, in this paper, the discrete form profile will represent exactly a cycloidal disk from a cycloidal reducer whose coordinates were determined using specific reverse engineering processes. Considering that the determination of the relative movements between the tool and the part, as well as the determination of the enveloping condition are relatively complicated and can be a source of errors, the purpose of this work is to highlight the advantages of the "virtual pole" method compared to the other specific methods of tool profiling that generate by enwrapping. At the same time, it is investigated if the "virtual pole" method can provide a support for the identification of the intermediate surface, with the help of graphic design environments, for profiling, in this case, the worm tools, since the "virtual pole" method cannot be applied as such in the case of disk-type tools, worm tools and cylindricalfrontal tools, due to the fact that the theorem is intended for the study of planar gearing. The advantages of using the "virtual pole" method are obvious compared to the classical tool profiling methods. The results demonstrated the accuracy of the method, by the fact that, using the "virtual pole" method, the calculation time for profiling the hob mill tool for generating a profile known in discrete form, namely the profile of the cycloidal disk, is significantly reduced. [ABSTRACT FROM AUTHOR]