Search

In this paper, a scalable end-to-end tree boosting system called XGBoost has been applied for predicting the magnitude of an earthquake from the early part of earthquake waveform data. This model uses the features extracted from the early P wave phase of the records as an input. The model's effectiveness has been verified by using data on earthquakes occurring in the Eurasian plate of Japan Islands from 1996 to 2021. Feature engineering has given 29 new features identified from the early P wave phase of the record, which show a high correlation with the magnitude of an earthquake. The comparison of predicted and actual magnitude shows that a trained XGboost model, which uses a single input record for magnitude prediction, gives an average prediction error of 0.004 ± 0.57 for earthquakes in the test dataset. In contrast, the average prediction error of –1.1 ± 0.80 and –0.65 ± 0.69 has been obtained for the magnitude estimated from conventional τc and Pd methods using the same test dataset. It is further seen that the average predicted magnitude of a single earthquake of magnitude 4.5 and 6.1 (MJMA) obtained by using multiple nearfield records using XGBoost model is 4.58 ± 0.33 and 6.32 ± 0.29, which is close to the actual magnitude of the earthquake. The results presented in this paper clearly show that the structured data can be effectively used by complex machine learning or deep learning models to predict earthquake magnitude from single or multiple records. [ABSTRACT FROM AUTHOR]

Due to the compact and complex internal structure of 110 kV three‐phase enclosed GIS disconnector, the voltage distribution between the contact breaks of each phase has an influence on the insulation performance of the GIS disconnector, and its internal insulation design is a difficult point to be solved. To solve this problem, this paper firstly establishes a three‐dimensional model of three‐phase enclosed GIS disconnecting switch and conducts a research on the distribution of electric field under a specific working condition, it is found that the outer corner of the static contact, the inner corner of the static arc contact, the inner corner of the outer end of the moving contact and the inner corner of the moving contact are the internal electric field concentration of the three‐phase enclosed GIS disconnector. Then, in order to further reduce the electric field strength and improve the insulation margin and performance, this paper proposes a method based on the combination of Extreme Learning Machine (ELM) and Multi‐objective Grasshopper Optimization Algorithm (MOGOA) to optimize the structural design of three‐phase enclosed GIS disconnector, and a ELM model is established, which takes the internal fillet radius of stationary arcing contact, the internal fillet radius of moving contact, the internal fillet radius of external end of dynamic and static contact, the radius of the outer corner of the two static side contacts and the opening distance of the grounding contact as the inputs, and takes the maximum electric field strength of the two static side contacts and the moving side contacts as outputs; Finally, the structural parameters are optimized by MOGOA algorithm. Compared to the pre‐optimization period, the electric field strengths of the bus bar static‐side contact, dynamic contact and static side contact of the B‐phase after optimization decreased from 20.2, 9.28 and 17.5 to 11.4, 6.65 and 11.5 kV/mm, respectively. The electric field strengths of the dynamic and static side contacts are significantly reduced, which is beneficial for improving the insulation performance of the three‐phase enclosed GIS disconnector. © 2023 Institute of Electrical Engineer of Japan and Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

Cirrhosis is a liver disease resulting from abnormal continuation of fibrosis, and ultrasound imaging is widely used for cirrhosis diagnosis because of its non‐invasiveness. However, due to unclear appearances of cirrhosis on ultrasound images, diagnoses are difficult and individual results possibly differ depending on the physician's experience. Recently, computer‐aided diagnostic systems using image processing and machine learning have been developed to help physicians detect cirrhosis as a 'Second opinion'. Some related studies have focused on a scenario where physicians set ROIs (Region of Interests) manually because selecting reliable ROIs for training a classifier and classification of patients is indispensable. But, the accuracy of such systems depends inherently on the quality of ROIs, and thus the workloads of physicians increase. In this paper, we propose a reliability evaluation method (REM) for each ROI based on its posterior probability and relationship to peripheral ROIs. The assumption of our proposal is that reliable regions of cirrhosis and normal can be observed in certain regions predominantly. We evaluated the effectiveness of the REM and its optimization for practical use. Experimental results showed that our proposed method curated reliable ROIs and improved classification performance in terms of AUC (Area Under the Curve). © 2024 Institute of Electrical Engineers of Japan and Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

Federated Learning (FL) is a distributed machine learning paradigm that enables multiple data holders to collaborate on building machine learning models while preserving the privacy of their data. FL can be categorized as horizontal or vertical, depending on the distribution characteristics of the data. Specifically, horizontal FL uses data partitioned in the sample space, whereas vertical FL uses data partitioned in the feature space. Traditional vertical FL methods aim to facilitate collaboration among clients to infer a single global target. However, these methods may be impractical because each client often has a unique target to be inferred. In this paper, we propose a novel vertical FL method, called personalized vertical federated learning (pvFed), which addresses this limitation by allowing each client to perform inferences specific to their individual task. To the best of our knowledge, no existing method currently addresses this limitation. The objective of pvFed is to construct a global model that generates a representation vector to support client inference. The global model, constructed using distillation and dimensionality reduction, takes a sample ID common to all clients as input and outputs a sample‐specific representation vector. Clients utilize the intermediate representation of their own model and the representation vectors output by the global model for inference. Because these vectors are not dependent on client‐specific tasks, clients can repurpose them for any additional tasks. Our experiments, conducted on two distinct data types—image and tabular data sets, under a vertical partitioning where each client had its own specific task, demonstrated the efficacy of vectors generated by the global model in pvFed for client inference. © 2024 Institute of Electrical Engineer of Japan and Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

Background: Bibliometric analysis is a mature method for quantitative evaluation of academic productivity. In view of the rapid development of research in the field of systemic lupus erythematosus (SLE) in the past decade, we used bibliometric methods to comprehensively analyze the literature in the field of SLE from 2013 to 2022. Methods: The relevant literature in the field of SLE from 2013 to 2022 was screened in the Web of Science Core Collection database. After obtaining and sorting out the data, CiteSpace and VOSviewer software were used to visualize the relevant data, and SPSS software was used for scientific statistics. Results: A total of 18,450 publications were included in this study. The number of articles published over the past 10 years has generally shown an upward trend, while Altmetric attention scores have also shown a clear upward trend in general and in most countries. Citation analysis and Altmetric analysis can mutually prove and supplement the influence of papers. The USA, China, Japan, Italy, and the UK are the most productive countries, but China and Japan are significantly inferior to other countries in terms of research influence. Four of the top ten authors are at the center of the collaboration network. LUPUS is the most contributing journal. The theme of systemic lupus erythematosus research mainly focuses on the pathogenesis, treatment, and management of SLE, and the emerging trend is related research on machine learning and immune cells. Conclusion: This study shows the research status of SLE, clarifies the main contributors in this field, discusses and analyzes the research hotspots and trends in this field, and provides reference for further research in this field to promote the development of SLE research. Key Points • Through bibliometric analysis, Altmetric analysis, and visual analysis, we reveal the global productivity characteristics of SLE-related papers in the past 10 years. • The number of global SLE-related studies has shown a significant increase, indicating that SLE is still a hot topic and deserves further study. • Citation analysis and Altmetric analysis can mutually prove and supplement the influence of papers, and the attention of related literature among non-professional researchers is increasing. • The theme of SLE research mainly focuses on the pathogenesis, treatment, and management of SLE. The emerging trend is machine learning and immune cells, which may provide new strategies for the diagnosis and treatment of SLE in the future. [ABSTRACT FROM AUTHOR]

Power line insulator defect identification usually suffers from complex backgrounds, small defect target sizes, and inconspicuous defect features. Traditional identification methods based on image processing, image analysis, and pattern classification have many limitations in solving the aforementioned problems. In recent decades, deep learning classification methods have gradually replaced traditional identification methods in the task of power line insulator defect identification. To accurately identify the locations of insulator defects, this paper proposes an insulator defect detection algorithm using an improved lightweight YOLOv4‐tiny network (ILYTN). First, the CBL (Conv‐BN‐LeakyReLU) modules of the backbone network are replaced with MobileViT blocks to enhance the feature extraction capability of the backbone network. Second, coordinate attention (CA) is introduced in the feature fusion part to improve the network's ability to focus on the location of defects. Finally, an EIOU (efficient intersection over union) loss function, instead of the original CIOU loss function, is used so that the convergence speed of the network can be improved. To verify the effectiveness of the proposed algorithm in this paper, the mViT‐yolo algorithm is compared with the mainstream Faster‐RCNN algorithm, SSD algorithm, YOLOv3 algorithm, and YOLOv4‐tiny algorithm. The experimental results show that the algorithm proposed in this paper outperforms all the above algorithms in terms of detection accuracy. Compared with the traditional YOLOv4‐tiny algorithm, the proposed algorithm increases the mean average precision (mAP) by 1.64%, the average precision (AP) of missing insulator defects by 0.32%, and the average precision (AP) of broken insulator defects by 4.96%. © 2023 The Authors. IEEJ Transactions on Electrical and Electronic Engineering published by Institute of Electrical Engineer of Japan and Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

Object tracking tasks often face problems such as illumination and appearance changes due to various scenarios. To effectively solve such problems, single object tracking algorithms in recent years have improved the stability of the tracker by introducing online learning, reducing tracking drift, and tracking objects stably even with large deformations. However, on the one hand, employing online learning increases computation costs, causing the tracker to run slower. On the other hand, because it learns from the tracking results, the tracking template will only be correctly updated if the object is accurately tracked. Any incorrect update will result in tracking failure. In this paper, we propose a filtering algorithm for online learning sample processing in object tracking to remedy these issues. Our method can select valid training samples and discard unnecessary training samples in object tracking. Numerous experiments demonstrate that our algorithm improves the tracking accuracy, reduces the computation complexity, and improves the running speed and run‐time. © 2023 Institute of Electrical Engineer of Japan and Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

Precise charging time prediction can effectively mitigate the inconvenience to drivers induced by inevitable charging behavior throughout trips. Although the effectiveness of the machine learning (ML) algorithm in predicting future outcomes has been established in a variety of applications (transportation sector), the investigation into electric vehicle (EV) charging time prediction is almost new. This calls for the investigation of the ML algorithm to predict EV charging time. The study developed an EV charging time prediction model based on two years of charging event data collected from 500 EVs in Japan. To predict EV charging time, this paper employed three ML algorithms: extreme learning machine (ELM), feed-forward neural network (FFNN), and support vector regression (SVR). Furthermore, ML algorithms parameters are optimized by a metaheuristic techniques: the gray wolf optimizer (GWO), particle swarm optimizer (PSO), and genetic algorithm (GA) to achieve higher accuracy and robustness. The prediction results reveal that GWO-based ML models yielded better results compared to other models. [ABSTRACT FROM AUTHOR]

In this paper, we explore the use of seismicity data on a single-station basis in site response characterization. We train a supervised deep-learning model, SeismAmp, to recognize and separate seismic site response with reference to seismological bedrock (VS  = 3.45 km s−1) in a broad frequency range (0.2–20 Hz) directly from single-station earthquake recordings (features) in Japan. Ground-truth data are homogeneously created using a classical multistation approach—generalized spectral inversion at a total number of 1725 sites. We demonstrate that site response can be reliably separated from single-station seismograms in an end-to-end approach. When SeismAmp is tested at new sites in both Japan (in-domain) and Europe (cross-domain), it achieves the lowest standard deviation among all tested single-station techniques. We also find that horizontal-to-vertical spectral ratio (HVSR) is not the optimal use of single-station recordings. The individual components of each record carry salient information on site response, especially at high frequencies. However, part of the information is lost in HVSR. SeismAmp could lead to improved site-specific earthquake hazard prediction in cases where recordings are available or can be collected at target sites. It is also a convenient tool to remove repeatable site effects from ground motions, which may benefit other applications, for example, improving the retrieval of seismic source parameters. Finally, SeismAmp is trained on data from Japan, future studies could explore transfer learning for practical applications in other regions. [ABSTRACT FROM AUTHOR]

With the continuous development of intelligent refrigerators, users are having higher and higher requirements for smart management of refrigerator contents. When using traditional machine learning algorithms to deal with refrigerator ingredients classification, problems such as slow recognition speed and low accuracy would often arise. At the same time, the classification accuracy is easily affected by the location of the content. Therefore, this paper uses continuous asymmetric convolutions and depthwise separable convolutions to improve the original Inception model and builds a classification prediction network based on the improved Inception model. The steps to applying the said model is as follows: First, we put the Radio Frequency Identification (RFID) tag on the food in the refrigerator, and then we collect the antenna data RSSI received by each tag through the reader. Finally, the RSSI is preprocessed and entered into the classification network to determine the location of the label. The proposed classification network has the advantages of possessing a simple structure, easy expansion of network depth, and fewer training parameters. The experimental results show that compared with other machine learning algorithms and pure convolution networks, the lightweight network model based on the improved Inception structure demonstrates great superiority. Compared with the original Inception structure, the enhanced network model is more lightweight. The number of parameters is reduced by 50%, and the amount of float‐point operations is reduced by 34%. Meanwhile, the accuracy of the lightweight model has been improved by 0.6%. Thus the improved model has better prediction performance in the classification of refrigerator content. © 2023 Institute of Electrical Engineer of Japan and Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

Peak ground acceleration (PGA) is a key parameter used in earthquake early warning systems to measure the ground motion strength and initiate emergency protocols at major projects. The traditional P -wave peak displacement-dependent PGA prediction model (Pd-PGA model) tends to underestimate the PGA for large earthquakes because it cannot make full use of the fault continuity rupture information hidden in the time-varying process of ground motion. In this paper, a continuous PGA prediction long short-term memory (LSTM) neural network model is proposed. The model takes eight sequential features of stations that are proxies of the energy and other physical parameters as input and provides the recorded PGA at the station as the target output. A total of 5961 records from 119 earthquakes recorded by the Japanese Strong-Motion Earthquake Network (K-NET) in Japan are used to train the neural network and 3433 records from 73 earthquakes are used as the test set to verify the model's generalization ability. The results show that within the same data set, the residuals of the predicted PGA for the proposed model are smaller than those of the Pd-PGA model and that the problem of PGA underestimation is resolved. The prediction accuracy also improves with increasing sequence length, which indicates that the LSTM neural network learns the rules hidden in the time series. To further verify the model's generalization ability, the model performance is analyzed for an M 7.3 earthquake that was not included in the training or test data sets. The results show that the residuals of the predicted PGA for the event are consistent with those for the test data set, indicating that the model has good generalization ability. [ABSTRACT FROM AUTHOR]

This paper proposes a micro‐Doppler radar‐based method to classify common pedestrians and pedestrians texting on a smartphone (called a smartphone zombies). We used a micro‐Doppler radar to collect motion data of the gait of participants texting on smartphones while walking, and then generated time‐frequency distribution images (spectrogram images). They were input into a convolutional neural network (CNN) to classify gait patterns. Results using measured data confirmed that the classification accuracy exceeded 90%, validating the effectiveness of the proposed method. Furthermore, by applying feature visualization by Grad‐Cam, we found that the motion of the leg swinging has essential features for the classification of smartphone zombies. © 2023 Institute of Electrical Engineer of Japan and Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

This paper compares the machine learning algorithm‐based prediction methods for the traction characteristics in linear traction induction motors operating at common working conditions, i.e. different slip and running velocity, with a symmetric or asymmetric secondary. These models provide a method for obtaining the dynamic characteristics in the motor that considers nonlinear effects. First, some analytical results for the prototype machine under different working conditions is calculated. Second, classification and feature extraction of traction characteristics results including thrust, transversal and vertical forces is made according to the different slip, running speed and lateral secondary displacement, and the results set is divided into training sets and test sets. Third, the prediction model established by different machine learning algorithms are analyzed and compared in principle. These algorithms in this paper mainly contain: artificial neural networks (ANNs), linear regression (LR), symbolic regression using GP, k‐Nearest Neighbour (kNN), random forests (RFRs). The machine learning algorithm‐based prediction methods are trained with the training set, and then the verified with the test set. Finally, this paper discusses the most optimal model for predicting traction characteristics. © 2021 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

Shikoku island, southwestern Japan lies in the western Nankai Trough and showcases along-strike segmentation of slow earthquake behavior. Whether the spatial variation of tremor behavior reflects the regional differences in structure/source properties and how much such differences can be recognized by the seismic signals themselves are two questions addressed in this paper. Taking advantage of advanced methods in recognizing and classifying signals using machine learning approaches, we attempt to answer them by conducting signal classification experiments in Shikoku. Based on the tremor catalog from 1 June 2014 to 31 March 2015, the tremors recorded in four different areas were treated as different classes and segmented into 60-s-long signals. The number of tremors in four different areas (A to D, from west to east) reached 15,000, 31,000, 10,000, and 16,000, respectively. To efficiently distinguish between tremors from different areas, we applied a k-nearest neighbor (k-NN) classifier with Fisher's class separability criteria to select the optimal feature subset. The resulting classification performance reached more than 90% at all 12 stations. We further designed a triangle test to select the features that can better represent the differences in source properties between areas. We found that the most efficient features were associated with (1) the number of peaks in the temporal evolution of discrete Fourier transforms and (2) the energy distribution in the autocorrelation function (ACF). To match the difference in behavior revealed by the ACF, the size of the tremor zone, which mainly controls how long the seismic energy lasts in a tremor episode, was determined to be largest in Area B and smallest in Area C. The heterogeneity of the asperities in a tremor zone, which may control how spiky the tremor signals developed over time, was determined to be strong in Areas B and C. Together with previously documented variations in slow earthquake behavior in the same area, we finally propose a conceptual model that provides a better understanding of the regional differences in the tremor source properties in Shikoku, Japan. [ABSTRACT FROM AUTHOR]

The shear wave velocity is among the key parameters that are responsible for damage caused during an earthquake. Determining shear wave velocity is a costly and time-consuming direct geophysical method. In the present paper, a machine learning model has been developed to obtain the subsurface shear wave velocity profile without using direct methods. The bore log information and the subsurface shear wave velocity profile available at various stations of Japan's Kyoshin network (K-NET) have been utilized for training this machine learning model. The parametric correlation study indicates that simple parameters like rock/soil type, the thickness of the layer in the model, and standard penetration test (SPT-N) value directly correlate with the medium's shear wave velocity. A stacked ensemble machine learning model named VelProfES (an acronym for Velocity Profiler using Ensemble machine learning models) has been developed in this paper and has been utilized for effective prediction of the shear wave velocity profile using basic information from borelog data. The dataset used in the training and testing of the machine learning model consists of borelog and shear wave velocity information from 1101 stations. Of 1101 stations, 657, 283, and 71 stations have been utilized for training, testing, and validating the machine learning model. Training, testing, and validation of the developed machine learning model consist of parameters from 12351, 5279, and 1388 velocity layers. The problem of data imbalance based on soil type has been addressed using an additional 10510 layers of synthetic borelog data generated from conditional generative adversarial networks (CTGANs). A feature and model ablation study was conducted for the VelProfES model to provide substantiation for the model and feature selection choices. The predicted shear wave velocity profiles were compared at specific stations, focusing on average velocities at 5, 10, 15, and 20 depths. Further, the predicted values have been compared with the empirical relation of Sil and Haloi (2017) and a trained polynomial model. The machine learning model demonstrates close alignment between predicted and actual values across a broad spectrum of velocities, a contrast not observed in the empirical relation and polynomial model. The results show that the machine learning models and augmented data generated using CTGANs can efficiently minimize the error between actual and predicted subsurface shear wave velocity values. • Describe detailed machine learning strategy for predicting shear wave velocity. • Generative Adversarial Network used to handle unbalanced data problem. • Describes correlation of shear wave velocity with Rock/soil type, layer, and SPT-N. • Comparison of described model with recently developed methods for all regions. • Performance of the machine learning model is predicted. [ABSTRACT FROM AUTHOR]

Machine learning (ML) has become a rising sophisticated technological application trend in the electrical industry in recent years. Such innovation provides optional methodologies for many existing applications, such as power and load profile forecasting, reliability evaluation, substation behavior detection and state observation of electrical equipment, and so on. This paper presents a review of various supervised and unsupervised ML techniques and applications for electrical power systems, including generation, transmission, distribution and micro‐grid. The algorithms and applications are mainly summarized from IEEE journals and the interest of this paper shows the roles and developments of most used algorithms and its corresponding extensions and performance in different applications. © 2021 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

Heave motion of ships is a complex nonlinear dynamic process and cannot be accurately forecasted using a single prediction model. In this paper, an effective combined forecasting method is proposed to perform ship's heave motion prediction. The proposed method combines back propagation neural network (BPNN), autoregressive model (AR) and extreme learning machine (ELM) through an induced ordered weighted averaging (IOWA) operator. The prediction accuracy is selected as the induced variable and the prediction results are sorted according to prediction accuracy and IOWA operator assigns larger weights to the position with the smallest prediction error. The optimal weights are determined by maximizing the B‐mode relational degree. Experimental results demonstrate its effectiveness of the proposed method. © 2022 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

Real estate price prediction takes an important role in the economy that can drive up and down the stock prices and even generate disruptive economic events. Many researchers have tried to understand the pricing mechanism with machine learning techniques such as support vector machine, neural network, random forest, and AdaBoost. The boundary problem, on the other hand, makes the pricing scheme more complicated, and this trend is accelerated especially in the situation of population decline in Japan. In this paper, we discuss how we could approach the boundary problem in real estate prediction. We propose a new comprehensive inference model extending and adapting XGBoost to the domain that has the boundary conditions problem by utilizing the distance between the instances in the domain data set to make the layers of bumpy boundaries smooth for more accurate predictions and robustness against the domain data set. The experiments result showed our proposed method performed well on both hypothetical data sets and actual real estate price data. © 2022 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

Recently, the whole world was hit by COVID‐19 pandemic that led to health emergency everywhere. During the peak of the early waves of the pandemic, medical and healthcare departments were overwhelmed by the number of COVID‐19 cases that exceeds their capacity. Therefore, new rules and techniques are urgently required to help in receiving, filtering and diagnosing patients. One of the decisive steps in the fight against COVID‐19 is the ability to detect patients early enough and selectively put them under special care. Symptoms of this disease can be observed in chest X‐rays. However, it is sometimes difficult and tricky to differentiate "only" pneumonia patients from COVID‐19 patients. Machine‐learning can be very helpful in carrying out this task. In this paper, we tackle the problem of COVID‐19 diagnostics following a data‐centric approach. For this purpose, we construct a diversified dataset of chest X‐ray images from publicly available datasets and by applying data augmentation techniques. Then, we employ a transfer learning approach based on a pre‐trained convolutional neural network (DenseNet‐169) to detect COVID‐19 in chest X‐ray images. In addition to that, we employ Gradient‐weighted Class Activation Mapping (GradCAM) to provide visual inspection and explanation of the predictions made by our deep learning model. The results were evaluated against various metrics such as sensitivity, specificity, Positive Predictive Value (PPV), Negative Predictive Value (NPV) and the confusion matrix. The resulting models has achieved an average detection accuracy close to 98.82%. © 2022 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

Ground motion prediction is an important and complex research subject in earthquake engineering and traditional approaches based on statistical regression have much room for improvement in prediction accuracy. Utilizing the 67,164 ground motion records from KiK-net and K-Net for 777 shallow crustal earthquakes between 1997 and 2019 in Japan, this paper proposes a ground motion prediction model XGBoost-SC based on the machine learning algorithm of eXtreme Gradient Boosting (XGBoost) for Japan. Magnitude, focal depth, hypo-central distance, V s30 , site altitude, and focal mechanism were used as the feature parameters, and XGBoost, Random Forest, and Deep Neural Networks (DNN) algorithms were selected for model training while Bayesian optimization was used to search for optimized hyperparameters to improve the prediction accuracy. XGBoost algorithm was selected for further study based on the comparison of results from the three algorithms. Residual change with magnitude and hypo-central distance, the probability distribution of residuals, residual standard deviation (σ), residual mean squared error (MSE), and Pearson correlation coefficient (R) were used as the evaluation parameters, and a comparison study was performed against the ground motion prediction equation based on traditional approaches. Actual earthquake events were selected to compare the prediction results against the observation records. To further validate and explain the proposed model, SHapley Additive exPlanations (SHAP) analysis was performed to explain the impact of selected feature parameters on the proposed model. The results demonstrate that the proposed XGBoost-SC model has good prediction stability for all periods, and its residual errors are smaller than those of other models. Therefore, the proposed model can better reflect the ground motion attenuation for shallow crustal earthquakes in Japan and can serve as a better model for ground motion prediction in future aseismic design and earthquake disaster mitigation efforts. • A Ground Motion Prediction Model was proposed for shallow crustal earthquakes. • The proposed model was more precise than traditional approaches. • Explained the effect of feature parameters on the prediction results. [ABSTRACT FROM AUTHOR]

In this paper, we address the task of segmenting small organs (i.e., the pancreas) from abdominal CT scans. As the target often occupies a relatively small region in the input image, deep neural networks can be easily confused by complex and variable backgrounds. We propose a method that uses a parallel multi‐scale network with an attention mechanism for pancreas segmentation, which can better grasp the balance between the semantic segmentation, classification, and localization tasks. We use a parallel network to connect the feature maps between different bottleneck layers, which contain rich semantic information and complete spatial information. We apply an attention module to enhance the key features of semantic information. Then, we fuse the two modules and apply the fused module as attention information on the feature map to ensure the full fusion between contextual semantic information and spatial information, thereby improving segmentation accuracy. We conduct extensive experiments on the NIH pancreas segmentation data set. In particular, our model achieves a mean coefficient Dice of 86.6. © 2021 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

Credit scoring model (CSM) is a risk management tool that assesses the credit worthiness of a customer borrower by estimating her probability of default based on historical data. Traditionally CSM is built by logit model or decision tree algorithm in financial companies, and in recent studies CSM has been integrated with machine learning algorithms such as random forest and gradient boosting to process a number of complex attributes of customer borrowers. On the other hand, CSM has been facing a critical challenge ‐ the domain adaptation of customer borrowers. For domain adaptation problem, transfer learning techniques are generally utilized, however, it is quite difficult to execute precise predictions for unknown domain datasets in CSM because the distributions of labels could be different depending on the characteristics of domains. Therefore, there is no appropriate transfer learning method to solve domain adaptation problem in credit scoring. In this paper we propose a comprehensive transfer learning framework using Bayesian network to extract useful knowledge based on probability distributions to predict probability of default of customer borrowers more precisely than existing machine learning and transfer learning methods. Experimental results showed the proposed method performed over the existing machine learning and transfer learning methods for accuracy of predictions. © 2021 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

Although measuring the ideal points of news media is essential for testing political communication theories based on spatial theory, prior methods of estimating ideal points of media outlets have various shortcomings, including high cost in terms of time and human resources and low applicability to different countries. We propose that unsupervised machine learning techniques for text data, specifically the combination of a text scaling method and latent topic modeling, can be applied to estimate ideal points of media outlets. We applied our proposed methods to editorial texts of ten national and regional newspapers in Japan, where prior approaches are not applicable because newspapers have never officially endorsed particular parties or candidates, and because high-quality training data for supervised learning are not available. Our two studies, one of which analyzed editorials on a single typically ideological topic while the other investigated all editorials published by the target papers in one year, confirmed the popular view of Japanese newspapers' ideological slant, which validates the effectiveness of our proposed approach. We also illustrate that our methods allow scholars to investigate which issues are closely related to the respective ideological positions of media outlets. Furthermore, we use the estimated ideal points of newspapers to show that Japanese people partially tend to read ideologically like-minded newspapers and follow such newspapers' Twitter accounts even though their slant is not explicit. [ABSTRACT FROM AUTHOR]

Facial expression is one of the most direct and effective ways to recognize emotions, widely used in human‐computer interaction, affective computing, and other research fields. Expression recognition can be divided into discrete expression classification and continuous dimensional emotion recognition. Most of the existing multi‐dimensional emotional estimation only considers the data under laboratory conditions. In this paper, facial emotion estimation is performed based on real‐world images and combined with the advantages of multi‐task learning and attention mechanism. We improve the multi‐task attention network (MTAN) from two aspects: task and feature. At the aspect of the task, the multi‐task collaborative attention network (MTCAN), which is based on task correlation, is proposed to solve task deviation in multi‐task learning. At the aspect of the feature, based on MTCAN, we came up with MTACN, which used the self‐attention mechanism to measure the importance of each attention module for each specific task. Then, we can capture the local‐to‐global connection in one step and fully exploit the feature within different levels of each task. Experimental results on the AffectNet dataset show that the performance of the model is significantly better than the original network, and the Root‐mean‐square error and consistency correlation coefficient results are superior to other existing models. © 2021 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

Aiming at the problem of unnecessary waste of power resources in the power system caused by the uncertainty of wind farm output, and then affecting the economics of the system. This paper proposes a power system scheduling model based on wind power output forecasting errors. In the forecasting stage, in view of the defect that the Fruit Fly Optimization Algorithm (FOA) is easy to fall into the local optimal value, an Adaptive Mutation Fruit Fly Optimization Algorithm (AMFOA) based on the flavor concentration variance is proposed. And then, the parameters of Extreme Learning Machine (ELM) are optimized by AMFOA to predict the wind power output. In the scheduling phase, a limit scenario is established based on the obtained prediction error to reduce the fluctuation of the wind farm output. When modeling, the influence of the prediction error is considered in the power balance equation, and it is constructed to maximize the probability of its establishment. And then, incorporate it into the objective function. Based on this, a scheduling model is established and solved by AMFOA. Finally, an example is used to verify the accuracy of the proposed forecasting model and the economics of the scheduling model. © 2021 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

Accurate and reliable fault location technology is essential to the stable operation of the modular multilevel converter‐high voltage direct current (MMC‐HVDC) system. Aiming at the difficulty of locating high‐resistance ground faults on MMC‐HVDC transmission lines, this paper proposes a method for fault location of transmission lines based on wavelet transform and deep belief network (DBN). First, the wavelet transform is used to decompose the original single pole ground fault voltage waveform, and then the high‐frequency and low‐frequency components obtained are used as training samples to train different DBN models, the final fault location results are obtained by superimposing the outputs of each model at last. The ±250 kV double‐ended MMC‐HVDC system model is established by using PSCAD/EMTDC, which can simulate the faults of different positions and different transition resistances. In order to verify the fault location performance of the proposed method, it is compared with two machine learning fault location methods. The results show that this method can accurately and reliably locate the single pole‐to‐ground fault of the transmission line with transition resistance of up to 10 000 Ω at low sampling frequency of 20 kHz. © 2020 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

This paper presents the performance and evaluation of a number of machine learning classifiers for the discrimination between the vegetation physiognomic classes using the satellite based time-series of the surface reflectance data. Discrimination of six vegetation physiognomic classes, Evergreen Coniferous Forest, Evergreen Broadleaf Forest, Deciduous Coniferous Forest, Deciduous Broadleaf Forest, Shrubs, and Herbs, was dealt with in the research. Rich-feature data were prepared from time-series of the satellite data for the discrimination and cross-validation of the vegetation physiognomic types using machine learning approach. A set of machine learning experiments comprised of a number of supervised classifiers with different model parameters was conducted to assess how the discrimination of vegetation physiognomic classes varies with classifiers, input features, and ground truth data size. The performance of each experiment was evaluated by using the 10-fold cross-validation method. Experiment using the Random Forests classifier provided highest overall accuracy (0.81) and kappa coefficient (0.78). However, accuracy metrics did not vary much with experiments. Accuracy metrics were found to be very sensitive to input features and size of ground truth data. The results obtained in the research are expected to be useful for improving the vegetation physiognomic mapping in Japan. [ABSTRACT FROM AUTHOR]

In many countries, and particularly in Japan, rising medical costs and shortfalls in the number of healthcare personnel are becoming serious national problems. This paper proposes a Preventive Care system with learning Communication robots ( PrevCareCom), aiming to provide a preventive care approach wherein the elderly can engage in exercise over the long term without getting bored while communicating with robots through match-up games. In the PrevCareCom, the preventive care exercises are based on a simple, traditional Japanese match-up game, and several kinds of robots are used as opponents. The proposed system encourages a sense of familiarity with the robots (agents) for elderly users based on the concept of human-agent interaction. Reinforcement learning methods are also used to adapt system parameters such as exercise intensity and robot behavioral policy to the athletic performance of each user. Several experiments were carried out with the cooperation of local governments and health and welfare facilities to investigate characteristics of the PrevCareCom and to evaluate how well it adapts to its users in terms of familiarity, quality and quantity of exercise. The results of experiments showed that the users' interest in the proposed system and sense of familiarity with the robots were encouraged by playing the game and interacting with the robots. Questionnaire results also showed that the PrevCareCom could provide appropriate exercise loads for the users' care prevention. [ABSTRACT FROM AUTHOR]

Storm surge is a genuine common fiasco coming from the ocean. Therefore, an exact forecast of surges is a vital assignment to dodge property misfortunes and to decrease a chance caused by tropical storm surge. Genetic Programming (GP) is an evolution-based model learning technique that can simultaneously find the functional form and the numeric coefficients for the model. Therefore, GP has been widely applied to build models for predictive problems. However, GP has seldom been applied to the problem of storm surge forecasting. In this paper, we propose a new method to use GP for evolving models for storm surge forecasting. Experimental results on datasets collected from the Tottori coast of Japan show that GP can evolve accurate storm surge forecasting models. Moreover, GP can automatically select relevant features when evolving storm surge forecasting models, and the models evolved by GP are interpretable. • The paper proposes a new approach to using Genetic Programming (GP) to evolve models for storm surge forecasting. • GP can evolve more accurate models for storm surge forecasting than other existing machine learning methods. • The model evolved by GP is more interpretable than models evolved by other (black-box) methods such as neural networks. • GP can automatically select relevant features when evolving storm surge forecasting models. [ABSTRACT FROM AUTHOR]

While Germany and Japan are going through major energy reforms, natural gas consumption is taking a growing share in their energy supply. This paper adopts a Machine Learning approach to assess the causal link between natural gas consumption and economic growth for both economies. A Causal Direction from Dependency (D2C) algorithm with the interconnection of the sub-class is employed using yearly data from 1970 to 2018. The interconnections of the sub-classes are found for both economies, indicating evidence of causalities operating in both directions. In addition, the propagation over the seven eras is linear and homogeneously continue for Japan, while this effect meets a stabilization phase in the fifth era for Germany. The empirical findings claim strong support for the existence of a bidirectional causality between these variables in Germany and Japan, which is in line with the "feedback hypothesis". Although the strength of this bidirectional relationship is clear for both economies, its time-propagation is expected to be longer for Japan. Accordingly, this study claims that the gas supply should be further strengthened to progressively replace the most polluting fuels (oil and coal) and ensure a feasible transition towards a renewable path. • The causal link between natural gas consumption and economic growth exists for both countries. • Predictive analysis highlights a propagation effect between natural gas consumption and economic growth for both economies. • We found empirical sustain for the "feedback hypothesis" for Germany and Japan. [ABSTRACT FROM AUTHOR]

Applications of machine learning on remote sensing data appear to be endless. Its use in damage identification for early response in the aftermath of a large-scale disaster has a specific issue. The collection of training data right after a disaster is costly, time-consuming, and many times impossible. This study analyzes a possible solution to the referred issue: the collection of training data from past disaster events to calibrate a discriminant function. Then the identification of affected areas in a current disaster can be performed in near real-time. The performance of a supervised machine learning classifier to learn from training data collected from the 2018 heavy rainfall at Okayama Prefecture, Japan, and to identify floods due to the typhoon Hagibis on 12 October 2019 at eastern Japan is reported in this paper. The results show a moderate agreement with flood maps provided by local governments and public institutions, and support the assumption that previous disaster information can be used to identify a current disaster in near-real time. [ABSTRACT FROM AUTHOR]

The development of new methods for estimating precise forest structure parameters is essential for the quantitative evaluation of forest resources. Conventional use of satellite image data, increasing use of terrestrial laser scanning (TLS), and emerging trends in the use of unmanned aerial systems (UASs) highlight the importance of modern technologies in the realm of forest observation. Each technology has different advantages, and this work seeks to incorporate multiple satellite, TLS- and UAS-based remote sensing data sets to improve the ability to estimate forest structure parameters. In this paper, two regression analysis approaches are considered for the estimation: random forest regression (RFR) and support vector regression (SVR). To collect the dependent variable, in situ measurements of individual tree parameters (tree height and diameter at breast height (DBH)) were taken in a Japanese cypress forest using the nondestructive TLS method, which scans the forest to obtain dense and accurate point clouds under the tree canopy. Based on the TLS data, the stem volume was then computed and treated as ground truth information. Topographic and UAS information was then used to calculate various remotely sensed explanatory variables, such as canopy size, canopy cover, and tree height. Canopy cover and canopy shapes were computed via the orthoimages derived from the UAS and watershed segmentation method, respectively. Tree height was computed by combining the digital surface model (DSM) from the UAS and the digital terrain model (DTM) from the TLS data. Topographic variables were computed from the DTM. The backscattering intensity in the satellite imagery was obtained based on L-band (Advanced Land Observing Satellite-2 (ALOS-2) Phased Array type L-band Synthetic Aperture Radar-2 (PALSAR-2)) and C-band (Sentinel-1) synthetic aperture radar (SAR). All satellite (10–25 m resolution), TLS (3.4 mm resolution) and UAS (2.3–4.6 cm resolution) data were then combined, and RFR and SVR were trained; the resulting predictive powers were then compared. The RFR method yielded fitting R2 up to 0.665 and RMSE up to 66.87 m3/ha (rRMSE = 11.95%) depending on the input variables (best result with canopy height, canopy size, canopy cover, and Sentinel-1 data), and the SVR method showed fitting R2 up to 0.519 and RMSE up to 80.12 m3/ha (rRMSE = 12.67%). The RFR outperformed the SVR method, which could delineate the relationship between the variables for better model accuracy. This work has demonstrated that incorporating various remote sensing data to satellite data, especially adding finer resolution data, can provide good estimates of forest parameters at a plot level (10 by 10 m), potentially allowing advancements in precision forestry. [ABSTRACT FROM AUTHOR]