Search

As one of the classic tasks in information retrieval, the core of image retrieval is to identify the images sharing similar features with a query image, aiming to enable users to find the required information from a large number of images conveniently. Street view image retrieval, in particular, finds extensive applications in many fields, such as improvements to navigation and mapping services, formulation of urban development planning scheme, and analysis of historical evolution of buildings. However, the intricate foreground and background details in street view images, coupled with a lack of attribute annotations, render it among the most challenging issues in practical applications. Current image retrieval research mainly uses the visual model that is completely dependent on the image visual features, and the multimodal learning model that necessitates additional data sources (e.g., annotated text). Yet, creating annotated datasets is expensive, and street view images, which contain a large amount of scene texts themselves, are often unannotated. Therefore, this paper proposes a deep unsupervised learning algorithm that combines visual and text features from image data for improving the accuracy of street view image retrieval. Specifically, we employ text detection algorithms to identify scene text, utilize the Pyramidal Histogram of Characters encoding predictor model to extract text information from images, deploy deep convolutional neural networks for visual feature extraction, and incorporate a contrastive learning module for image retrieval. Upon testing across three street view image datasets, the results demonstrate that our model holds certain advantages over the state‐of‐the‐art multimodal models pre‐trained on extensive datasets, characterized by fewer parameters and lower floating point operations. Code and data are available at https://github.com/nwuSY/svtRetrieval. [ABSTRACT FROM AUTHOR]

Development of a reliable and intelligent traffic monitoring system is highly desired to improve the transportation safety and establish future transportation plans due to the fast growth of vehicle population. Vehicle classification is one of the most critical subsystems where existing ones suffer from privacy concerns, requirements of complicated systems, and high maintenance cost. This paper reports a novel vehicle classification method by utilizing a triboelectric sensor to accurately identify vehicles. Novelty of this method originates from using triboelectric sensor and machine learning method with important advantages over current alternatives by providing an easy installation, simple operation, noninvasive measurement, cost-effective manufacturing, and highly accurate classification. To make a classification, vehicle toys' signals were acquired from triboelectric sensor and then applied to a deep learning algorithm. The 1932 sensor output data were grouped into a set of seven vehicle toys with different wheelbases, and number of tires passing on are used to train and optimize 1D-CNN model. The utilized 1D-CNN model achieved accuracy, f1-score, precision, and recall as 96.38%, 0.9638, 0.9658, and 0.9637, respectively. [ABSTRACT FROM AUTHOR]

The heating of high-voltage isolating switches above 110 kV will cause serious safety accidents. To identify the heating faults of isolating switches accurately, this study uses natural language processing technology, knowledge graph (KG) technology and machine learning algorithm to mine the structured and unstructured mixed data. Firstly, based on bidirectional encoder representations (BERT) pre-training model, a bidirectional long short-term memory network and conditional random field model are introduced to identify entities related to high-voltage isolating switches and the thermal failures from the text data. Further, a convolutional neural network fusion attention mechanism model is leveraged to recognize the corresponding entity relationships from the text records. Then, the KGs of isolating switches and the heating faults are constructed, and a path ranking algorithm is proposed to deduce and identify the possible factors causing overheating on the KGs. Combined with the structured data, a support vector machine model based on focal loss function (FL-SVM) is proposed to recognize the heating faults of isolating switches. Experiments show that the performances of all six models used in this paper on the hybrid dataset are higher than those on the structured dataset. Such results suggest that joint text-structured data mining can compensate for insufficient structured data and thus, achieving a better heating failure recognition, especially for the infrequent malfunctions. Meanwhile, the proposed FL-SVM performs better than other models due to its excellent ability to solve the problem of imbalanced faults and normal samples. [ABSTRACT FROM AUTHOR]

Background: Glaucoma, the second leading cause of global blindness, demands timely detection due to its asymptomatic progression. This paper introduces an advanced computerized system, integrates Machine Learning (ML), convolutional neural networks (CNNs), and image processing for accurate glaucoma detection using medical imaging data, surpassing prior research efforts. Method: Developing a hybrid glaucoma detection framework using CNNs (ResNet50, VGG-16) and Random Forest. Models analyze pre-processed retinal images independently, and post-processing rules combine predictions for an overall glaucoma impact assessment. Result: The hybrid framework achieves a significant 95.41% accuracy, with precision and recall at 99.37% and 88.37%, respectively. The F1 score, balancing precision and recall, reaches a commendable 93.52%. These results highlight the robustness and effectiveness of the hybrid framework in accurate glaucoma diagnosis. Conclusion: In summary, our research presents an innovative hybrid framework combining CNNs and traditional ML models for glaucoma detection. Using ResNet50, VGG-16, and Random Forest in an ensemble approach yields remarkable accuracy, precision, recall, and F1 score. These results showcase the methodology's potential to enhance glaucoma diagnosis, emphasizing its promising role in early detection and preventing irreversible vision loss. The integration of ML and DNNs in medical imaging analysis suggests a valuable path for future advancements in ophthalmic healthcare. [ABSTRACT FROM AUTHOR]

Classifying imbalanced data presents a significant challenge, and many studies have proposed methodologies to address this issue. Among them, Convolutional Neural Networks have demonstrated superior performance for imbalanced image classification. This paper initially employs various data pre-processing methods such as over-sampling, under-sampling, and SMOTE to enhance the original dataset. Subsequently, an Ensemble CNN learning model is used to train and predict the data. In order to comprehensively evaluate models trained on imbalanced data, we used metrics such as Accuracy, Recall, Precision, F1-score, and G-mean. On the CIFAR-10 and Fashion-MNIST datasets, different samples from each category were extracted as imbalanced data for experimental research. Compared to the AdaBoost-DenseNet model, our proposed methodology increases the test accuracy on the CIFAR-10 dataset by 9%. Similarly, the F1-score and G-mean improved by 0.096 and 0.069, respectively. Compared to traditional methodologies, our proposed method significantly improves accuracy, recall, precision, and other performance indicators. [ABSTRACT FROM AUTHOR]

Aiming at the problem that the cement production process is inherently affected by uncertainty, time delay, and strong coupling among variables, this paper proposed a novel soft sensor of free calcium oxide in a cement clinker. The model utilizes a dual-parallel integrated structure with an optimized integration of one-dimensional convolutional neural networks, long and short-term memory networks, graphical neural networks, and extreme gradient boosting. The proposed model can mitigate the risks associated with overfitting while incorporating the strengths of each individual model and excels in extracting both local and global features as well as temporal and spatial characteristics from the original time series data, ensuring its stability. The experimental results demonstrate that this dual-parallel integrated model exhibits superior robustness, predictive accuracy, and generalization capabilities when compared to single models or enhancements made to other deep learning algorithms. [ABSTRACT FROM AUTHOR]

To predict the response of masonry buildings to various types of loads, engineers use finite element models, specifically solid-element and macro-element models. For predicting masonry responses to seismic events in particular, equivalent frame models—a subcategory of macro-element models—are a common choice because of their low computational cost. However, an existing bottleneck in modeling pipelines is generating the geometry of the model, which is currently a slow and laborious process that is done manually using computer-aided design tools. In this paper, we address this by automating the modelling process using recent advancements in computer vision and machine learning. We present an image-based end-to-end pipeline that automatically generates finite element meshes for solid-element and equivalent-frame models of the outer walls of free-standing historical masonry buildings. As the input, our framework requires RGB images of the buildings that are processed using structure-from-motion algorithms, which create 3D geometries, and convolutional neural networks, which segment the openings and their corners. These layers are then combined to generate level of detail models. We tested our pipeline on structures with irregular surface geometries and opening layouts. While generating the solid element mesh from the level of detail model is straightforward, generating equivalent frame models required algorithms for segmenting the façade and the meshing. Experts in the field analyzed the generated equivalent frame models and determined them to be useful for numerical modeling. These finite element geometries will be invaluable for future predictions of the seismic response of damaged and undamaged buildings. The codes and dataset are publicly available for future studies and benchmarking (https://github.com/eesd-epfl/FEM%5fbuildings and https://doi.org/10.5281/zenodo.8094306). [ABSTRACT FROM AUTHOR]

Bats play a pivotal role in maintaining ecological balance, and studying their behaviors offers vital insights into environmental health and aids in conservation efforts. Determining the presence of various bat species in an environment is essential for many bat studies. Specialized audio sensors can be used to record bat echolocation calls that can then be used to identify bat species. However, the complexity of bat calls presents a significant challenge, necessitating expert analysis and extensive time for accurate interpretation. Recent advances in neural networks can help identify bat species automatically from their echolocation calls. Such neural networks can be integrated into a complete end-to-end system that leverages recent internet of things (IoT) technologies with long-range, low-powered communication protocols to implement automated acoustical monitoring. This paper presents the design and implementation of such a system that uses a tiny neural network for interpreting sensor data derived from bat echolocation signals. A highly compact convolutional neural network (CNN) model was developed that demonstrated excellent performance in bat species identification, achieving an F1-score of 0.9578 and an accuracy rate of 97.5%. The neural network was deployed, and its performance was evaluated on various alternative edge devices, including the NVIDIA Jetson Nano and Google Coral. [ABSTRACT FROM AUTHOR]

As the number of electronic gadgets in our daily lives is increasing and most of them require some kind of human interaction, this demands innovative, convenient input methods. There are limitations to state-of-the-art (SotA) ultrasound-based hand gesture recognition (HGR) systems in terms of robustness and accuracy. This research presents a novel machine learning (ML)-based end-to-end solution for hand gesture recognition with low-cost micro-electromechanical (MEMS) system ultrasonic transducers. In contrast to prior methods, our ML model processes the raw echo samples directly instead of using pre-processed data. Consequently, the processing flow presented in this work leaves it to the ML model to extract the important information from the echo data. The success of this approach is demonstrated as follows. Four MEMS ultrasonic transducers are placed in three different geometrical arrangements. For each arrangement, different types of ML models are optimized and benchmarked on datasets acquired with the presented custom hardware (HW): convolutional neural networks (CNNs), gated recurrent units (GRUs), long short-term memory (LSTM), vision transformer (ViT), and cross-attention multi-scale vision transformer (CrossViT). The three last-mentioned ML models reached more than 88% accuracy. The most important innovation described in this research paper is that we were able to demonstrate that little pre-processing is necessary to obtain high accuracy in ultrasonic HGR for several arrangements of cost-effective and low-power MEMS ultrasonic transducer arrays. Even the computationally intensive Fourier transform can be omitted. The presented approach is further compared to HGR systems using other sensor types such as vision, WiFi, radar, and state-of-the-art ultrasound-based HGR systems. Direct processing of the sensor signals by a compact model makes ultrasonic hand gesture recognition a true low-cost and power-efficient input method. [ABSTRACT FROM AUTHOR]

In recent years, the rapid advancement of artificial intelligence (AI) technology has made remarkable strides in the music industry, profoundly impacting various aspects of our lives. This progress has extended to revolutionizing industries, streamlining processes, and enhancing personalization in music. AI's expanding capabilities have driven computer vision, natural language processing, and machine learning breakthroughs. As music enthusiasts increasingly turn to online platforms, the need for effective recommendation systems has become more pressing. The exponential growth of musical content poses a challenge in discovering music tailored to individual tastes. This paper proposed a personalized music recommendation system powered by deep learning, explicitly utilizing convolutional neural networks (CNNs) to address the challenges posed by expanding music libraries and diverse song resources. In today's era of ever-growing music databases and increasing content diversity, more than traditional retrieval methods are needed to efficiently connect users with their preferred track. Traditional music platforms rely solely on retrieval mechanisms, necessitating manual inputs such as artist names and titles to locate specific songs. However, this approach needs in-depth music data analysis and delivers personalized recommendations generated by various deep learning-based algorithms. This paper bridges this gap by harnessing deep learning algorithms for comprehensive music data analysis and personalized song suggestions. The proposed CNN-based recommendation system explores the domains of deep learning and content-based approaches, empowering users to discover their favorite tunes swiftly. This innovative mechanism leverages CNNs and deep learning algorithms to offer efficient, personalized, and data-driven music recommendations, transforming the user experience with their music collections. In experimental results, the proposed CNN model is compared with various deep learning algorithms, including DNN, RNN, and LSTM, as well as traditional algorithms like SVM, KNN, RF, and DT. Impressively, the CNN model achieved the highest accuracy at 95%, surpassing all other algorithms. [ABSTRACT FROM AUTHOR]

This paper constructs a dataset of handwritten Indus signs employing a social approach. A writing system called the Indus script was created in the Indus civilization. It has been decoded numerous times throughout the years, but it has not yet been fully deciphered. Due to a lack of information and the scarcity of evidence, the mystery of the Indus signs has not yet been fully solved. Recently, there has been an increase in demand for huge datasets in order to use cutting-edge machine learning techniques. Considering the restricted availability of images of authentic Indus signs, this paper proposes creating an Indus signs dataset by asking participants to draw the Indus signs while referring to the image of the original Indus signs. A web application was developed and used to collect the 44 participants' handwritten images of ten Indus signs. To show the availability of the constructed dataset, it is used to train convolutional neural networks. The experimental result demonstrates that the model can classify the images of original Indus script with 70% accuracy. [ABSTRACT FROM AUTHOR]

Deep Learning (DL) and Machine Learning (ML) are effectively utilized in various complicated challenges in healthcare, industry, and academia. The Internet of Drones (IoD) has lately cropped up due to high adjustability to a broad range of unpredictable circumstances. In addition, Unmanned Aerial Vehicles (UAVs) could be utilized efficiently in a multitude of scenarios, including rescue missions and search, farming, mission-critical services, surveillance systems, and so on, owing to technical and realistic benefits such as low movement, the capacity to lengthen wireless coverage zones, and the ability to attain places unreachable to human beings. In many studies, IoD and UAV are utilized interchangeably. Besides, drones enhance the efficiency aspects of various network topologies, including delay, throughput, interconnectivity, and dependability. Nonetheless, the deployment of drone systems raises various challenges relating to the inherent unpredictability of the wireless medium, the high mobility degrees, and the battery life that could result in rapid topological changes. In this paper, the IoD is originally explained in terms of potential applications and comparative operational scenarios. Then, we classify ML in the IoD-UAV world according to its applications, including resource management, surveillance and monitoring, object detection, power control, energy management, mobility management, and security management. This research aims to supply the readers with a better understanding of (1) the fundamentals of IoD/UAV, (2) the most recent developments and breakthroughs in this field, (3) the benefits and drawbacks of existing methods, and (4) areas that need further investigation and consideration. The resultssuggest that the Convolutional Neural Networks (CNN) method is the most often employed ML method in publications. According to research, most papers are on resource and mobility management. Most articles have focused on enhancing only one parameter, with the accuracy parameter receiving the most attention. Also, Python is the most commonly used language in papers, accounting for 90% of the time. Also, in 2021, it has the most papers published. [ABSTRACT FROM AUTHOR]

In recent years, the rapid advancement of artificial intelligence (AI) has revolutionized various industries, with image processing playing an important role in the revolution. AI-powered image analysis has opened up new possibilities in healthcare diagnostics, autonomous navigation, and security surveillance, fuelling demand for sophisticated image-processing solutions. This paper describes an innovative approach to AI image processing that combines multi-sensor data fusion with convolutional neural networks (CNNs) within a fuzzy neural network framework. This integration uses data from various sensors, including cameras, lidar, and radar, to improve the robustness and precision of image analysis and interpretation. The T–S model serves as the foundation for the information fusion strategy. A comprehensive investigation of deep learning algorithms reveals inherent strengths such as robustness and parallelism. However, it also identifies limitations, particularly in image segmentation tasks, characterized by challenges like premature convergence and prolonged computation times. The paper proposes a quantization technique for deep learning algorithms to address these issues and introduces chaotic optimization to expedite convergence rates. It also presents a novel three-dimensional Otsu threshold segmentation method based on CNNs, which overcomes noise susceptibility in traditional two-dimensional approaches. Integrating Gray morphology and this three-dimensional Otsu threshold segmentation method results in the development of a three-dimensional Gray Otsu model. This model is the basis for designing a fitness function for the CNN algorithm, optimizing its efficacy. Experimental validation demonstrates the proposed algorithm's effectiveness, achieving an impressive 91% accuracy rate while displaying robust noise resistance and versatility. Comparative assessments against other leading AI architectures, including multilayer perceptron, radial basis function network, recurrent neural network, and long short-term memory network, affirm the superior performance achieved by the proposed approach. [ABSTRACT FROM AUTHOR]

In order to better evaluate and promote human health, this paper analyzes the influence of different inertial-measurement-unit signals, different sensor locations, different activity intensities and different signal fusion schemes on the accuracy of physical strength consumption estimation during walking and running activities. Different pattern recognition methods, such as the Counts-based linear regression model, the typical non-linear model based on decision tree and artificial neural network, and the end-to-end convolutional neural network model, are analyzed and compared. Our findings are as follows: 1) For the locations of sensors during walking and running activities, the physical strength consumption prediction accuracy at the ankle location is higher than that at the hip location. Therefore, wearing an inertial-measurement-unit at the ankle can improve the accuracy of the model. 2) Regarding the types of activity signals during walking and running activities, the impact of accelerometer signals on hip and ankle prediction accuracy is not significantly different, while the gyroscope model is more sensitive to the location, with higher prediction accuracy at the ankle than at the hip. In addition, the physical strength consumption prediction accuracy of accelerometer signals is higher than that of gyroscope signals, and fusion of accelerometer and gyroscope signals can improve the accuracy of physical strength consumption prediction. 3) For different data analysis models during walking and running activities, the artificial neural network model that integrates different sensor locations and inertial-measurement-unit signals with different activity intensities has the lowest mean squared error for the measurement of physical strength consumption. The non-linear models based on decision tree and artificial neural network have better physical strength consumption prediction capabilities than the Counts-based linear regression model, especially for high-intensity activity energy consumption prediction. In addition, feature engineering models are generally better than convolutional neural network model in terms of overall performance and prediction results under the three different activity intensities. Furthermore, as the activity intensity increases, the performance of all physical strength consumption calculation models decreases. We recommend using the artificial neural network model based on multi-signal fusion to estimate physical strength consumption during walking and running activities because this model exhibits strong generalization ability in cross-validation and test results, and its stability under different activity intensities is better than that of the other three models. To the best of our knowledge, this paper is the first to delve deeply and in detail into methods for estimating physical strength consumption. Undoubtedly, our paper will have an impact on research related to topics such as intelligent wearable devices and subsequent methods for estimating physical strength consumption, which are directly related to physical health. [ABSTRACT FROM AUTHOR]

Code smells violate software development principles that make the software more prone to errors and changes. Researchers have developed code smell detectors using manual and semi-automatic methods to identify these issues. However, three key challenges have limited the practical use of these detectors: developers' subjective perceptions of code smells, lack of consensus in the detection process, and difficulty in setting appropriate detection thresholds. While code smell detection using machine learning has progressed significantly, there still appears to be a gap in understanding the effective utilization of deep learning (DL) approaches. This paper aims to review and identify current methods for code smell detection using DL techniques. A systematic literature review is conducted on 35 primary studies from a collection of 8739 publications between 2013 and the present. The analysis reveals that common code smells detected include Feature Envy, God Classes, Long Methods, Complex Classes, and Large Classes. The most popular DL algorithms used are Recurrent Neural Networks (RNN) and Convolutional Neural Networks (CNN), often combined with other techniques for better results. Algorithms that train models on large datasets with fewer independent variables demonstrate exemplary performance. The paper also highlights open issues and provides guidelines for future metric identification and selection research. [ABSTRACT FROM AUTHOR]

Construction sites remain one of the most hazardous workplaces globally. To improve workplace safety in the construction industry and reduce the personal injuries and socioeconomic impacts resulting from workplace accidents, tacit knowledge containing fundamental causes of accidents or specific contextual factors can be extracted from past accident narrative reports. However, manually analyzing unstructured or semistructured textual data stored in records is a daunting task, and requires the use of automated and intelligent technologies to achieve rapid and accurate knowledge acquisition. Therefore, this paper proposes a text self-classification model based on deep learning natural language processing (NLP) technology for automated classification of construction site accident cases by accident type. First, combined with two statistical measures, mutual information and information entropy, the preprocessed text data were subjected to phrase segmentation to identify more complete and accurate accident precursor information without human intervention. Then a complete multilayer and multisize convolutional neural network (CNN) model was constructed using pretrained Word2Vec word embeddings for text self-classification tasks. Finally, the test results of the CNN classification algorithm were compared with the practical application results of three shallow learning algorithms, and the performance of different types of classification algorithms was evaluated. The results showed that the CNN-based deep learning algorithm developed in this paper demonstrated excellent feature extraction and learning abilities in the task of automatic text classification in the field of NLP. This not only demonstrated that reliable accident prevention knowledge could be obtained from the textual descriptions of construction accidents, but also provided a novel model reference for document archiving and information retrieval. [ABSTRACT FROM AUTHOR]

Partial discharge (PD) is a common issue in power transformers that can lead to catastrophic failures if left undetected. Time reversal (TR) is a well-known technique in signal processing that can reconstruct signals by reversing the direction of time. The paper investigates the use of time reversal and the integration of time reversal with convolution neural networks (CNNs) for diagnosing PD in power transformers. We compare the performance of these techniques on a dataset of PD signals collected from power transformers. We propose a novel method of using time reversal as a pre-processing step to improve the accuracy of CNNs on noisy or distorted signals. Our experimental results demonstrate that this approach can significantly enhance the performance of CNNs on various datasets, including speech, audio, and image datasets. This paper provides a novel approach to signal processing and demonstrates the potential of time reversal as a pre-processing step in CNNs. [ABSTRACT FROM AUTHOR]

Simple Summary: The pepper [Capsicum annuum L. (Solanales: Solanaceae)] is one of the most economically important vegetable crops and the demand for peppers has been increasing. In the process of pepper crop growth, the types and quantity of insect pests are increasing, which has become the first natural biological disaster threatening crop yield and quality. In particular, the green peach aphid [Myzus persicae Sulzer (Hemiptera: Aphididae)] is one of the most threatening insect pests in pepper cultivation, which causes a lot of damage, such as necrosis, wilting, chlorosis, and defoliation. Hence, accurately recognizing and counting the aphids is essential for ensuring the excellent productivity of pepper crops. Due to the time-consuming and inefficient work of traditional methods, there have been deep learning methods for crop pest recognition and counting in recent years. The YOLO algorithm is one of the most effective deep learning algorithms used for object detection. To investigate aphid recognition and counting of pepper crop growth, this paper develops an improved YOLOv5 approach based on a deep convolutional neural network (CNN) in the climate chamber environment. Experimental results in the aphid dataset indicate that the proposed method can achieve significant improvements in terms of recognition accuracy and counting performance. Due to changes in light intensity, varying degrees of aphid aggregation, and small scales in the climate chamber environment, accurately identifying and counting aphids remains a challenge. In this paper, an improved YOLOv5 aphid detection model based on CNN is proposed to address aphid recognition and counting. First, to reduce the overfitting problem of insufficient data, the proposed YOLOv5 model uses an image enhancement method combining Mosaic and GridMask to expand the aphid dataset. Second, a convolutional block attention mechanism (CBAM) is proposed in the backbone layer to improve the recognition accuracy of aphid small targets. Subsequently, the feature fusion method of bi-directional feature pyramid network (BiFPN) is employed to enhance the YOLOv5 neck, further improving the recognition accuracy and speed of aphids; in addition, a Transformer structure is introduced in front of the detection head to investigate the impact of aphid aggregation and light intensity on recognition accuracy. Experiments have shown that, through the fusion of the proposed methods, the model recognition accuracy and recall rate can reach 99.1%, the value mAP@0.5 can reach 99.3%, and the inference time can reach 9.4 ms, which is significantly better than other YOLO series networks. Moreover, it has strong robustness in actual recognition tasks and can provide a reference for pest prevention and control in climate chambers. [ABSTRACT FROM AUTHOR]

Some architects struggle to choose the best form of how the building meets the ground and may benefit from a suggestion based on precedents. This paper presents a novel proof of concept workflow that enables machine learning (ML) to automatically classify three-dimensional (3D) prototypes with respect to formulating the most appropriate building/ground relationship. Here, ML, a branch of artificial intelligence (AI), can ascertain the most appropriate relationship from a set of examples provided by trained architects. Moreover, the system classifies 3D prototypes of architectural precedent models based on a topological graph instead of 2D images. The system takes advantage of two primary technologies. The first is a software library that enhances the representation of 3D models through non-manifold topology (Topologic). The second is an end-to-end deep graph convolutional neural network (DGCNN). The experimental workflow in this paper consists of two stages. First, a generative simulation system for a 3D prototype of architectural precedents created a large synthetic database of building/ground relationships with numerous topological variations. This geometrical model then underwent conversion into semantically rich topological dual graphs. Second, the prototype architectural graphs were imported to the DGCNN model for graph classification. While using a unique data set prevents direct comparison, our experiments have shown that the proposed workflow achieves highly accurate results that align with DGCNN's performance on benchmark graphs. This research demonstrates the potential of AI to help designers identify the topology of architectural solutions and place them within the most relevant architectural canons. [ABSTRACT FROM AUTHOR]

Artificial intelligence is changing and influencing our world. As one of the main algorithms in the field of artificial intelligence, convolutional neural networks (CNNs) have developed rapidly in recent years. Especially after the emergence of NASNet, CNNs have gradually pushed the idea of AutoML to the public's attention, and large numbers of new structures designed by automatic searches are appearing. These networks are usually based on reinforcement learning and evolutionary learning algorithms. However, sometimes, the blocks of these networks are complex, and there is no small model for simpler tasks. Therefore, this paper proposes POSS-CNN aiming at target recognition and detection, which employs a multi-branch CNN structure with PSNC and a method of automatic parallel selection for super parameters based on a multi-branch CNN structure. Moreover, POSS-CNN can be broken up. By choosing a single branch or the combination of two branches as the "benchmark", as well as the overall POSS-CNN, we can achieve seven models with different precision and operations. The test accuracy of POSS-CNN for a recognition task tested on a CIFAR10 dataset can reach 86.4%, which is equivalent to AlexNet and VggNet, but the operation and parameters of the whole model in this paper are 45.9% and 45.8% of AlexNet, and 29.5% and 29.4% of VggNet. The mAP of POSS-CNN for a detection task tested on the LSVH dataset is 45.8, inferior to the 62.3 of YOLOv3. However, compared with YOLOv3, the operation and parameters of the model in this paper are reduced by 57.4% and 15.6%, respectively. After being accelerated by WRA, POSS-CNN for a detection task tested on an LSVH dataset can achieve 27 fps, and the energy efficiency is 0.42 J/f, which is 5 times and 96.6 times better than GPU 2080Ti in performance and energy efficiency, respectively. [ABSTRACT FROM AUTHOR]

A pipeline is the most efficient way to transport water from one place to another. Due to aging, corrosion, and external factors, the pipeline is prone to damage, which causes leaks. Many machine learning (ML) and deep learning (DL) methods are available to address this issue. This paper does an experimental study on available methods in ML and DL for leak detection for the collected data using an acousto-optic sensor. The experimental setup comprises of an acousto-optic sensor made of an erbium-doped fiber laser (EDFL), galvanized iron pipeline, a tank, a pump, and a data acquisition unit. The dimensions of the galvanized pipeline looped with the water tank are a length of 40 m, an inner diameter of 89 mm, and an outer diameter of 90 mm. The diameter of the simulated leak aperture is 5 mm. The methods analyzed in this study are k-means, k-medoids, Naive Bayes (NB), support vector machines (SVM), k-nearest neighbor (KNN), decision tree (DT), categorical boosting (CatBoost), random forest (RF), XGBoost, AdaBoost, and one-dimensional convolutional neural network (1DCNN). ML algorithms need a feature extraction technique because the data collected from the experiment is too large and contains redundant information. Feature extraction reduces the data size by extracting essential information. This paper extracts ten features from raw data. Among the ML algorithms, AdaBoost gives the highest prediction accuracy of 98.02%. This paper also implements eight models of 1DCNN, and Model 1 shows the best prediction accuracy of 98.16%, which is the highest compared with all the other classifiers in ML and DL for one-dimensional time series acousto-optic sensor data. [ABSTRACT FROM AUTHOR]

Photovoltaic (PV) fault detection is crucial because undetected PV faults can lead to significant energy losses, with some cases experiencing losses of up to 10%. The efficiency of PV systems depends upon the reliable detection and diagnosis of faults. The integration of Artificial Intelligence (AI) techniques has been a growing trend in addressing these issues. The goal of this systematic review is to offer a comprehensive overview of the recent advancements in AI-based methodologies for PV fault detection, consolidating the key findings from 31 research papers. An initial pool of 142 papers were identified, from which 31 were selected for in-depth review following the PRISMA guidelines. The title, objective, methods, and findings of each paper were analyzed, with a focus on machine learning (ML) and deep learning (DL) approaches. ML and DL are particularly suitable for PV fault detection because of their capacity to process and analyze large amounts of data to identify complex patterns and anomalies. This study identified several AI techniques used for fault detection in PV systems, ranging from classical ML methods like k-nearest neighbor (KNN) and random forest to more advanced deep learning models such as Convolutional Neural Networks (CNNs). Quantum circuits and infrared imagery were also explored as potential solutions. The analysis found that DL models, in general, outperformed traditional ML models in accuracy and efficiency. This study shows that AI methodologies have evolved and been increasingly applied in PV fault detection. The integration of AI in PV fault detection offers high accuracy and effectiveness. After reviewing these studies, we proposed an Artificial Neural Network (ANN)-based method for PV fault detection and classification. [ABSTRACT FROM AUTHOR]

At present, the packing method of "plastic bag–buffer packing–packing paper box" is adopted for bearing packaging. However, the common packing method has a poor packing effect and poor versatility. In this study, a new biomimetic cobweb cushion is proposed to solve the problem of insufficient cushioning capacity of high-precision bearing cushion packaging pads. First, according to the nature of cobweb form, the cobweb cushion structure configuration is determined. Next, based on the structure of the cushion and the relationship between the parameters of radial thread and spiral thread, a mechanical and target optimization model is established. The stress nephogram of bearing and the cobweb cushion are analyzed under three drop heights of 381, 610, and 700 mm, in the finite element simulation software to ensure that the maximum bearings stress is not beyond the material yield strength. Via the 3D printing technology, a cobweb cushion shell cast is made. Drop tests of the bearing were performed, and the results were verified with the finite element simulation analysis. This research can provide technical support for the protection of high-precision bearings from accidental drops during transportation. [ABSTRACT FROM AUTHOR]

In areas with high tuberculosis (TB) prevalence, high mortality rate has significantly increased over the past few decades. Even though tuberculosis can be treated, areas with high disease burden continue to have insufficient screening tools, leading to diagnostic delays and incorrect diagnoses. As a result of these challenges, a computer-aided diagnostics (CAD) system has been developed that can automatically detect tuberculosis. There are few different methods that can be used to screen for tuberculosis; however, chest X-ray (CXR) is most commonly used and strongly suggested because it is so effective in identifying lung irregularities. Over past ten years, we have seen a meteoric rise in amount of research conducted into application of machine learning strategies to examination of chest X-ray images for screening regarding pulmonary abnormalities. Particularly, we have also noticed significant interest in testing for TB. This attentiveness has increased in tandem with phenomenal progress that has been made in deep learning (DL), which is predominately founded on convolutional neural networks (CNNs). Because of these advancements, significant research contributions have been made in field of DL techniques for TB screening by utilizing CXR images. The main focus of this paper is to emphasize favorable methods and data collection, as well as methodological contributions, identify data collections, and identify challenges. [ABSTRACT FROM AUTHOR]

Deep learning integration in cancer diagnosis enhances accuracy and diagnosis speed which helps clinical decision-making and improves health outcomes. Despite all these benefits in cancer diagnosis, the present AI models in urology cancer diagnosis have not been sufficiently reviewed systematically. This paper reviews the artificial intelligence approaches used in cancer diagnosis, prediction, and treatment of urology cancer. AI models and their applications in urology subspecialties are evaluated and discussed. The Scopus, Microsoft Academic and PubMed/MEDLINE databases were searched in November 2022 using the terms "artificial intelligence", "neural network", "machine learning," or "deep learning" combined with the phrase "urology cancers". The search was limited to publications published within the previous 20 years to identify cutting-edge deep-learning applications published in English. Irrelevant review articles and publications were eliminated. The included research involves two kinds of research analysis: quantitative and qualitative. 48 articles were included in this survey. 25 studies proposed several approaches for prostate cancers, while 15 were for bladder cancers. 8 studies discussed renal cell carcinoma and kidney cancer. The models presented to detect urology cancers have achieved high detection accuracy (77–95%). Deep learning approaches that use convolutional neural networks have achieved the highest accuracy among other techniques. Although it is still progressing, the development of AI models for urology cancer detection, prediction, and therapy has shown significant promise. Additional research is required to employ more extensive, higher-quality, and more recent datasets to the clinical performance of the proposed AI models in urology cancer applications. [ABSTRACT FROM AUTHOR]

Nowadays, machine learning (ML) has attained a high level of achievement in many contexts. Considering the significance of ML in medical and bioinformatics owing to its accuracy, many investigators discussed multiple solutions for developing the function of medical and bioinformatics challenges using deep learning (DL) techniques. The importance of DL in Internet of Things (IoT)-based bio- and medical informatics lies in its ability to analyze and interpret large amounts of complex and diverse data in real time, providing insights that can improve healthcare outcomes and increase efficiency in the healthcare industry. Several applications of DL in IoT-based bio- and medical informatics include diagnosis, treatment recommendation, clinical decision support, image analysis, wearable monitoring, and drug discovery. The review aims to comprehensively evaluate and synthesize the existing body of the literature on applying deep learning in the intersection of the IoT with bio- and medical informatics. In this paper, we categorized the most cutting-edge DL solutions for medical and bioinformatics issues into five categories based on the DL technique utilized: convolutional neural network, recurrent neural network, generative adversarial network, multilayer perception, and hybrid methods. A systematic literature review was applied to study each one in terms of effective properties, like the main idea, benefits, drawbacks, methods, simulation environment, and datasets. After that, cutting-edge research on DL approaches and applications for bioinformatics concerns was emphasized. In addition, several challenges that contributed to DL implementation for medical and bioinformatics have been addressed, which are predicted to motivate more studies to develop medical and bioinformatics research progressively. According to the findings, most articles are evaluated using features like accuracy, sensitivity, specificity, F-score, latency, adaptability, and scalability. [ABSTRACT FROM AUTHOR]

This paper proposes a forward layer-wise learning algorithm for CNNs in classification problems. The algorithm utilizes the Separation Index (SI) as a supervised complexity measure to evaluate and train each layer in a forward manner. The proposed method explains that gradually increasing the SI through layers reduces the input data's uncertainties and disturbances, achieving a better feature space representation. Hence, by approximating the SI with a variant of local triplet loss at each layer, a gradient-based learning algorithm is suggested to maximize it. Inspired by the NGRAD (Neural Gradient Representation by Activity Differences) hypothesis, the proposed algorithm operates in a forward manner without explicit error information from the last layer. The algorithm's performance is evaluated on image classification tasks using VGG16, VGG19, AlexNet, and LeNet architectures with CIFAR-10, CIFAR-100, Raabin-WBC, and Fashion-MNIST datasets. Additionally, the experiments are applied to text classification tasks using the DBPedia and AG's News datasets. The results demonstrate that the proposed layer-wise learning algorithm outperforms state-of-the-art methods in accuracy and time complexity. [ABSTRACT FROM AUTHOR]

To bring more intelligence to edge systems, Federated Learning (FL) is proposed to provide a privacy-preserving mechanism to train a globally shared model by utilizing a massive amount of user-generated data on devices. FL enables multiple clients collaboratively train a machine learning model while keeping the raw training data local. When the dataset is horizontally partitioned, existing FL algorithms can aggregate CNN models received from decentralized clients. But, it cannot be applied to the scenario where the dataset is vertically partitioned. This manuscript showcases the task of image classification in the vertical FL settings in which participants hold incomplete image pieces of all samples, individually. To this end, the paper discusses AdptVFedConv to tackle this issue and achieves the CNN models' aim for training without revealing raw data. Unlike conventional FL algorithms for sharing model parameters in every communication iteration, AdptVFedConv enables hidden feature representations. Each client fine-tunes a local feature extractor and transmits the extracted feature representations to the backend machine. A classifier model is trained with concatenated feature representations as input and ground truth labels as output at the server-side. Furthermore, we put forward the model transfer method and replication padding tricks to improve final performance. Extensive experiments demonstrate that the accuracy of AdptVFedConv is close to the centralized model. [ABSTRACT FROM AUTHOR]

It is of great significance to accurately grasp ground-level ozone concentration on a wide scale in order to cope with the increasingly serious ground-level ozone pollution problem. Currently, satellite technology can be used to monitor ground-level ozone concentration on a wide scale, but there is the problem of insufficient spatial resolution. The use of ground monitoring stations to detect ground-level ozone concentration can obtain more accurate ozone concentration around the station, but there are problems of limited number of stations and sparse distribution. Therefore, it is still a challenge to accurately grasp the ground-level ozone concentration at a large scale and high spatial resolution. To address this issue, this paper proposes a ground-level ozone concentration inversion algorithm based on deep learning methods, incorporating data from remote sensing satellites, ground monitoring stations, meteorological conditions, geography, and land use. Taking Yunnan Province in China as the experimental region and utilizing five-fold cross-validation, the model's Mean Square Error (MSE) reached 340.45 and Root Mean Square Error (RMSE) reached 18.45. Compared to machine learning models, the error decreased by 14.53%, outperforming Random Forest models, Convolution Neural Networks, and Vision Transformers. This study offers a referential approach for accurately capturing ground-level ozone concentrations at a large scale with high spatial resolution. [ABSTRACT FROM AUTHOR]

In a Multiple-Input Multiple-Output Orthogonal Frequency Division Multiplexing (MIMO-OFDM) method, multiple antennas can be used on either the transmitter or receiver end to improve the system capacity, data throughput, and robustness. OFDM has been used as the modulation system that divides the data stream into multiple parallel low-rate subcarriers. MIMO enhances the system by utilizing spatial diversity and multiplexing abilities. Modulation classification in the MIMO-OFDM systems describes the process of recognizing the modulation scheme used by the communicated signals in a MIMO-OFDM communication system. This is a vital step in receiver design as it enables proper demodulation of the received signals. In this paper, an Enhanced Modulation Classification Approach using an Arithmetic Optimization Algorithm with Deep Learning (EMCA-AOADL) is developed for MIMO-OFDM systems. The goal of the presented EMCAAOADL technique is to detect and classify different types of modulation signals that exist in MIMO-OFDM systems. To accomplish this, the EMCA-AOADL technique performs a feature extraction process based on the Sevcik Fractal Dimension (SFD). For modulation classification, the EMCA-AOADL technique uses a Convolution Neural Network with Long Short-Term Memory (CNN-LSTM) approach. Finally, the hyperparameter values of the CNN-LSTM algorithm can be chosen by using AOA. To highlight the better recognition result of the EMCA-AOADL approach, a comprehensive range of simulations was performed. The simulation values illustrate the better results of the EMCA-AOADL algorithm. [ABSTRACT FROM AUTHOR]

Adverse weather conditions, such as snow-covered roads, represent a challenge for autonomous vehicle research. This is particularly challenging as it might cause misalignment between the longitudinal axis of the vehicle and the actual direction of travel. In this paper, we extend previous work in the field of autonomous vehicles on snow-covered roads and present a novel approach for side-slip angle estimation that combines perception with a hybrid artificial neural network pushing the prediction horizon beyond existing approaches. We exploited the feature extraction capabilities of convolutional neural networks and the dynamic time series relationship learning capabilities of gated recurrent units and combined them with a motion model to estimate the side-slip angle. Subsequently, we evaluated the model using the 3DCoAutoSim simulation platform, where we designed a suitable simulation environment with snowfall, friction, and car tracks in snow. The results revealed that our approach outperforms the baseline model for prediction horizons ⩾ 2 seconds. This extended prediction horizon has practical implications, by providing drivers and autonomous systems with more time to make informed decisions, thereby enhancing road safety. [ABSTRACT FROM AUTHOR]

One of the predominant health issues affecting Saudi Arabia and leading to many complications is Type 2 diabetes (T2D). Early detection and significant preventative measures lead to curbing and controlling the health issue. There are fewer datasets in the literature for the detection of T2D in the Saudi population. Past studies using Saudi data have favoured machine learning algorithms to classify T2D. Although the application of this data in machine learning is evident, no studies exist in the literature that compare this data, especially those related to deep learning algorithms. This study's objective is to use specific Saudi data to develop multiple deep learning models that could be used to detect T2D. The research uses a Deep Neural Network (DNN), an Autoencoder (AE), and a Convolutional Neural Network (CNN) to create predictive models and compare their performance with a traditional machine learning classifier used on the same dataset that outperformed other machine learning algorithms such as a Decision Forest (DF). Various metrics were used to evaluate the effectiveness of the models, such as accuracy, precision, recall, F1 score and area under the ROC curve (AUC) where the ROC acts as a receiver operating characteristic curve. There are two cases in this paper: (i) uses all features of the dataset and (ii) uses six of the ten features, such as DF. In case (i), the results were shown that AE outperformed other models with the highest accuracy for imbalanced and balanced data 81.12% and 79.16%, respectively. The results for case (ii) showed that AE scored the highest 81.01% accuracy with imbalanced data compared to DF and DF achieved the highest accuracy of 82.1% with balanced data. As a result, both cases explored in this study revealed that AE has a constant superior performance if imbalanced data is used. In contrast, DF demonstrated the highest accuracy when a balanced dataset was used with a feature set reduction. They help to identify the undiagnosed T2D, and they are essential for professionals in Saudi Arabia in the health sector to promote health connections, identify risks and contain or improve their diabetes management. [ABSTRACT FROM AUTHOR]

Detection and classification methods for the melt pool state in laser direct energy deposition (L-DED) can significantly help predict defects and mechanical properties of L-DED metal parts. Although traditional machine learning algorithms based on physical modeling methods and convolutional neural networks have recently been introduced into melt pool state identification, these methods rely on complex artificially designed features or cannot simultaneously detect defects in multiple dimensions. In this paper, a novel bilateral stream neural network was designed for melt pool identification, which performs defect identification in two label dimensions simultaneously. Two sets of single-channel experiments were designed to collect the dataset captured by a high-speed camera. By cutting the metal parts and marking them with professional equipment operated by professionals, the dataset was labeled according to the bonding condition and dilution rate criteria. Without an additive model structure, the model achieved 95.2% accuracy in identifying defects in the bonding condition and 92.8% in determining deficiencies in the dilution rate. In order to explain the identification mechanism of the model, the CAM method was utilized for the visual display of the model recognition process, which provides a potential application solution for the online monitoring method of the L-DED. [ABSTRACT FROM AUTHOR]

Parkinson disease (PD) is a type of neurodegenerative disorder that affects the motor movement of the patient. But each technique has its own advantages or disadvantages. In gene, speech and handwriting data model, the feature extraction and reduction is an important step for efficient classification. These two steps require proper attention for selection and also require high processing time as compared to other data model like images. Because in image modality, the deep learning algorithm can be applied that can perform all process and automate the classification. As compared to these domains, the signal produces better and best results. Because the electroencephalogram (EEG) signal are taken from the brain using electrodes and it helps to observe the brain signals effectively and immediately as compared to the other data modals. Hence, in this paper, the wavelet transform will be used to decompose the signals and statistical features will be extracted from the transformed signal. Here, the satin bower bird optimization will be used for both type of wavelet selection and feature reduction process for final classification. The reduced feature set will be classified using Ensemble Neural Network type including InceptionV3, DenseNet, MobileNet, Xception, and NasNet) recently proposed for medical image classification. The whole process will be realized using MATLAB R2021a software and its performance will be evaluated in terms of Accuracy and is compared against Automated Tunable Q-wavelet transform performance. The proposed ensemble method, employing EEG signal processing and neural networks, achieved a 97% success rate in discriminating PD datasets, surpassing Convolutional Neural Network (CNN) and Machine Learning (ML) classifications (88% –92%). Utilizing MATLAB R2021a, its superiority over Q-wavelet transform was evident, signifying improved PD dataset discrimination. [ABSTRACT FROM AUTHOR]

This paper introduces an AI-based approach to detect human-made objects and changes in these on land parcels. To this end, we used binary image classification performed by a convolutional neural network. Binary classification requires the selection of a decision boundary, and we provided a deterministic method for this selection. Furthermore, we varied different parameters to improve the performance of our approach, leading to a true positive rate of 91.3% and a true negative rate of 63.0%. A specific application of our work supports the administration of agricultural land parcels eligible for subsidiaries. As a result of our findings, authorities could reduce the effort involved in the detection of human made changes by approximately 50%. [ABSTRACT FROM AUTHOR]

In this paper, a novel method is introduced to diagnose the cardiac arrhythmias. Automatic diagnosis of cardiac arrhythmias is crucial for successful treatment of heart diseases, and nowadays, machine learning is used to facilitate this purpose. For accurate classification of arrhythmias, it is vital to extract some appropriate features which distinguish different classes. In this research, a deep convolutional neural network is used to extract the features. Since the heart rates of various patients are very different, arrhythmias of a specific class will be placed in different locations of feature space. To reduce the intra-class variations, each patient's heart rate is mapped with a dedicated function to increase its similarity to the heart rate of a training patient. The proposed patient-specific mapping decreases the intra-class variation and significantly increases the classification accuracy of cardiac arrhythmias. The calculated accuracy is about 91.08%, which shows the better performance compared to other similar works. [ABSTRACT FROM AUTHOR]

Deep learning methods have shown significant potential in computer vision applications within the field of civil engineering. One important area of research is the detection and classification of construction machines using convolutional neural networks (CNNs). However, a major challenge in adopting machine learning models for vision systems lies in hyperparameter optimization. Therefore, the objective of this paper is to propose a meticulous statistical approach to hyperparameter tuning for deep learning in the visual classification of construction machines. It is noteworthy that the method for selecting hyperparameters utilizes statistical concepts, including analysis of variance and the Tukey test. Moreover, three research questions were formulated to address statistical differences in performance when varying the CNN architecture, optimizer and configuration (architecture and optimizer). Two additional research questions were posed to determine the recommended hyperparameters for the task and assess performance differences for non-selected hyperparameters. For this purpose, 18 hyperparameter combinations were analyzed, encompassing three CNN architectures (DenseNet, Mobilenet and VGG16) and six optimizers (adadelta, adagrad, rmsprop, adam, adamax and sgd). The Alberta Construction Image Dataset, consisting of 2850 images of three construction machines (excavator, dump truck and concrete mixer truck), was used in the experiments. The results demonstrate that the hyperparameter configurations exhibit statistically significant differences in image classification. Furthermore, the recommended combination (DenseNet × adagrad) yielded the highest average accuracy results in both binary tests (90.0%) and multiclass classification (77.8%). [ABSTRACT FROM AUTHOR]

This paper provides a method for improving the photovoltaic conversion efficiency and optical attributes of silicon solar cells manufactured from as‐cut boron doped p‐type multi‐crystalline silicon wafers using acid‐based chemical texturization via machine learning. A decreased reflectance, which can be attained by the right chemical etching conditions, is one of the key elements for raising solar cell efficiency. In this work, the mc‐Silicon wafer surface reflectance is obtained under (<2%) after optimization of wet chemical etching. The HF + HNO3 + CH3COOH chemical etchant is used in the ratio 1:3:2 at different conditions of the etching duration of 1 min, 2 min, 3 min, and 4 min, respectively. The as‐cut boron doped p‐type mc‐silicon wafers are analysed with ultraviolet–visible spectroscopy, optical microscopy, Fourier transforms infrared spectroscopy, thickness profilometer, and scanning electron microscopy before and after etching. The chemical etching solution produces good results in 3 min etched wafer, with a reflectivity value of <2%.The reflectivity and optical images are inputs to the convolutional neural network model and the linear regression model to obtain the etching rate for better reflectivity. The classification model provides 99.6% accuracy and the regression model results in the minimum mean squared error (MSE) of 0.062. [ABSTRACT FROM AUTHOR]