Search

In recent years, the object‐tracking algorithm based on Siamese network has gradually become the mainstream algorithm in the field of object tracking due to its characteristics of balancing speed and accuracy. The majority of Siamese‐based trackers only use the first frame extraction template for subsequent tracking in order to prevent the introduction of noise. However, merely with a single initial template employed, it is difficult to achieve the best performance of the tracker in the face of complex tracking environments such as occlusion, motion blur, and non‐rigid deformation. Therefore, the present paper proposes a new multi‐template fusion module based on graph attention network (G‐M module), which consists of two parts: a graph‐attention‐network‐based feature‐embedding module (G module) and a multi‐template fusion module (M module). It can greatly reduce the background noise introduced by template updating while improving the tracker's ability to adapt to changes in object appearance. In addition, in order to maximize the value of G‐M module, the present paper also puts forward a two‐stage template update threshold judgment mechanism. The Pearson correlation coefficient (PCCs) is introduced and combined with APCE and the maximum response value (F‐max) to filter out reliable templates for updating. In this paper, the proposed method is applied to the SiamFC and SiamFC++ trackers. Extensive experiments on mainstream data sets, such as OTB2015, VOT2016, and GOT‐10 k, show that the proposed method can effectively update the tracking template and improve the tracker performance. © 2022 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

Influenced by cultural background, economic development, social system, education system, and other factors, there is still a big gap between Chinese institutions and developed countries in flute teaching, even with our neighbors, South Korea and Japan. Under the influence of cultural background, economic development, social system, and educational system, there is still a very big gap between Chinese colleges and universities and developed countries in flute teaching, even with our neighbors, South Korea and Japan. Because of its local perception and weight-sharing structure, the convolutional neural network is closer to the biological neural network in the real world. The weight-sharing structure reduces the complexity of the neural network, which can avoid the complexity of feature extraction and classification process in data reconstruction. This paper studies the analysis and optimization of flute playing and teaching system based on a convolutional neural network. By applying local perception field and parameter sharing in a convolutional neural network at the same time and adding multiple filters, it can not only effectively reduce the number of parameters but also extract features layer by layer. In the process of convolution, the parameters of the characteristic map obtained by each layer decrease layer by layer, but the number increases gradually. Based on the analysis of the problems faced by the flute performance teaching, this paper puts forward the corresponding solutions in order to promote the flute performance teaching in China to achieve better results. [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]

Most face detection and recognition tasks are based on the training of intact facial images and corresponding labels. Both the three‐dimensional (3D) structure and two‐dimensional (2D) appearance from the frontal view of human faces are bilaterally symmetrical in general. However, sometimes, illumination on the left and right halves of faces is uneven. In such cases, the symmetrical characteristic of human faces can facilitate expressing distinct identity information. This is because even if one side of the facial image is corrupted by noise, the opposite side can still be used for feature extraction. This paper proposes an automatic selection of the better half of the face using only a half‐face for identity recognition. Unlike the MegaFace challenge of recognizing millions of identities in the wild, this paper focuses on building recognition systems for a small number of people with fewer training images; the recognition system can, for example, build access control systems for research laboratory members or family members. This paper proposes an artificial face image construction method and a half‐face training strategy for transfer learning of pretrained conventional neural network models. Extensive experimental results show that the proposed method improves the performance of state‐of‐the‐art models by utilizing the symmetrical characteristics of human faces. © 2020 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

This paper investigates a deep learning‐based wind‐forecasting model to establish an accurate forecasting model which can support the increasing growth of wind power generation. The wind forecasting means wind speed and direction forecasting at the same time. Proposed forecasting model consists of three‐dimensional convolutional neural network and deep convolutional long short‐term memory (3DCNN‐DConvLSTM), and forecasts the wind vector which expressed as time‐sequential images. DConvLSTM model learns spatiotemporal features from time‐series image data that represent a spatial and temporal change of wind speed and direction. The forecasting model combined of 3DCNN and DConvLSTM is effective to decrease training time, and forecasting error in comparison to the DConvLSTM model. Input of the forecasting model is wind speed and direction that is expressed as an image on 2D coordinate and uses the measured data by the Automated Meteorological Data Acquisition System (AMeDAS), Japan. Forecasting accuracy with one‐hour ahead and its usefulness of the proposed forecasting model is evaluated with simulation results for four seasons that is typical of Japan climate, and demonstrated by comparison with fully connected‐LSTM (FC‐LSTM), encoder‐decoder based 3DCNN (ED‐3DCNN), DConvLSTM, and persistent models. © 2022 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

Electric vehicles (EV) are the future of mobility solutions. The EV's are driven by an electric motor with the help of a power electronic interface. The power electronic interface needs to be designed in an efficient way both in mechanical and electrical aspects. This paper proposes the concept of design, simulation and analysis of a 10 kW Multi‐Device Interleaved DC‐DC Boost Converter (MDIBC) to drive a 4 kW Induction Motor. The output DC voltage is to be maintained at a constant 400 V irrespective of the input variations. MDIBC consists of semi‐controlled switches topology excited by Phase Shifted PWM technique to reduce the ripple current in interleaving inductors. The dual loop control of the MDIBC is implemented using both PI and ANN controllers. The MATLAB/Simulink environment is used to simulate and analyze the design and control implementations. The results obtained through the implementation of ANN control proves that the control of MDIBC is simpler and is superior when compared to PI control. © 2022 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

Summary: This paper proposes a framework for estimating tensions of stay cables of cable‐stayed bridge with and without lateral attachments under limited information on cable properties. For stay cables without lateral attachments, tension is estimated using three known parameters, namely, cable length, mass per unit length, and measured frequency while still accounting for unknown features like cable bending stiffness, axial stiffness, cable inclination, and restraint boundary conditions at the cable ends. The methodology is then extended to stay cables with lateral attachments (dampers and cross ties), whose properties are also considered as unknown parameters. The framework is proposed via the application of back‐propagation artificial neural networks (ANNs). The finite difference formulation of the cable model is derived to create datasets for training, validation, and testing in the ANNs scheme. The feasibility and robustness of the proposed framework were confirmed through numerical verifications. The results indicated that tension in cables without lateral attachments was successfully evaluated using at least two measured frequencies, whereas cables with lateral attachments needed at least three measured frequencies to achieve high accuracy. Finally, the proposed framework was applied to estimate tensions in stay cables of Tatara Bridge, the longest cable‐stayed bridge in Japan. The results demonstrated that the proposed method could estimate the cable tension with acceptable accuracy. [ABSTRACT FROM AUTHOR]

In order to provide guidance for demand‐side management and improve energy efficiency, the accuracy of residential electricity demand forecasting plays a significant role. Data‐driven methods and deep learning network methods have been proved an effective methods for time series forecasting. In the context, the current research work proposes a novel neural network model based on convolutional neural network (CNN)‐attention‐bidirectional long‐short term memory (BiLSTM) to predict residential energy consumption. The proposed approach combines CNN, attention mechanism, and BiLSTM, initially implementing CNN to extract the effective features of the original data. Subsequently, the extracted segments are employed as the input of attention‐BiLSTM to predict energy consumption, where the attention mechanism comprehensively considers the output of BiLSTM neurons and assigns weights to neurons at each timestamp. In this article, we evaluate the proposed method using household electricity consumption data. This paper selects different input timestamp lengths (10/60/120 min, respectively) to validate the model's performance. We compare evaluation indicators (root mean square error, mean absolute error, mean absolute percentage error [MAPE]) between the proposed method and the state‐of‐the‐art prediction methods. The experimental result shows that the proposed method achieves higher energy consumption forecasting accuracy and has the lowest average MAPE (3.7%). © 2022 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

This paper presents a neural network model to estimate wave spectra from wave parameters by the zero-crossing analysis, maximum, significant, and average wave heights and their corresponding wave periods. The present model was trained with the wave data in the JES (Japan/East Sea), and then applied to wave measurements in the Yellow Sea and the South Sea for its feasibility at other seas. According to the present study, the predicted wave spectra well agree with the measurements, even the wave data not in model training. This comparison shows that the accuracy of the present model is comparable to that of JONSWAP wave spectrum. Furthermore, the present model was employed to the wave hindcasting to see its applicability, while JONSWAP wave spectrum also being applied. The significant wave heights from both numerical simulations well agree with the measured ones. However, the average RMSE of significant wave period with the present neural network was 7.08 % smaller than the other, and its Pearson's correlation coefficient is 6.33 % larger on average. From this, it can be seen that the present model was well applied to the wave hindcasting without considering the local wave measurements. • This paper presents a neural network model to estimate wave spectra from wave parameters by the zero-crossing analysis, maximum, significant, and average heights and the corresponding wave periods. • The wave spectra predicted by the present model well agree with the measured ones. The accuracy of present model is comparable to that of JONSWAP wave spectrum with Hwang et al.'s (2021) peak enhancement factor, comparing them. • The present model was employed to the wave hindcasting in Young-il bay to see its applicability. The numerical simulation with incident waves by JONSWAP wave spectrum was also carried out to compare the applicability of both models. • The numerical results show that the significant wave heights from both models are similar to each other. However, the significant wave period with the present model is more accurate than the other. [ABSTRACT FROM AUTHOR]

This paper presents a deep neural network‐based technique to recognize pets' body movements. The motivation to develop the technique arises from the fact that there are an increasing number of animal lovers, say dog or cat lovers, who spend a tremendous amount of effort to take care of their beloved pets, and even to capture the pets' images in daily life and at particular memorable moments for sharing among friends. For illustrative purposes, the recognition model was trained and then tested using a number of cats' images. As it turned out, the model well recognized three body movements: eating, tail raising and yawn, with a recognition accuracy up to 99.45%. Using this recognition technique, pets' images can be captured automatically once specific movements are detected, and job as a pet photographer can be made easy accordingly. © 2021 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

The mobile mapping system (MMS) can acquire dense point-clouds of roads and roadside features. Roads are often separated into roadways and walkways in many urban areas. Since guardrails are installed to separate roadways and sidewalks, it is important to detect guardrails from point-clouds and reconstruct their 3D models for 3D street maps. Since there are a large variety of designs for guardrails in Japan, flexible methods are required for detection and reconstruction of guardrails. In this paper, we propose a new method for extracting guardrails from point-clouds, and reconstructing their 3D models. Since the MMS captures point-clouds and camera images synchronously, guardrails are detected using both point-clouds and images. In our method, point-clouds are segmented into small segments, and corresponding images are cropped from camera images. Then cropped images are classified into two classes of guardrails and others using the convolutional neural network. When guardrail points are obtained, 3D models of guardrails are reconstructed. However, point-clouds of guardrails are too sparse to reconstruct 3D shapes when guardrails consist of thin pipes. Since the same unit shape repeatedly appears in a guardrail, we create dense point-clouds by superimposing points of unit shapes. Then we reconstruct 3D shapes of pipes, beams, and poles of guardrails. In our evaluation using point-clouds in urban areas, our method could achieve good results of extraction and shape reconstruction of guardrails. [ABSTRACT FROM AUTHOR]

This paper presents a reservoir computing technique using Asynchronous Block‐Based Neural Networks (ABBNN). Echo State Networks and Liquid State Machines introduced a new paradigm in artificial neural networks (ANN), known as Reservoir Computing (RC). A recurrent neural network (RNN) in Reservoir Computing is generated randomly, and only a readout is trained. The reservoir computing greatly facilitated the practical RNN application and outperformed classical RNN in many tasks. ABBNN is an extended version of the classical block‐based neural network model. ABBNN, an evolvable neural network model, provides a model‐free estimation of nonlinear dynamical systems. To propose a hardware‐aware model of reservoir computing and high‐speed training, we introduce Block‐Based Reservoir Computing (BBRC). BBRC provides a flexible architecture. The architecture based on the basic blocks provides a regularity that supports scalable hardware. In contrast, BBRC also supports randomness based on the internal configuration. Both are significant features in achieving an optimal hardware reservoir computer. © 2021 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

Purpose – The purpose of this paper is to design an innovative performance modeling system by jointly using data envelopment analysis (DEA) and artificial neural network (ANN). The hybrid DEA-ANN model integrates performance measurement and prediction frameworks and serves as an adaptive decision support tool in pursuit of best performance benchmarking and stepwise improvement. Design/methodology/approach – Advantages of combining DEA and ANN methods into an optimal performance prediction model are explored. DEA is used as a preprocessor to measure relative performance of decision-making units (DMUs) and to generate test inputs for subsequent ANN prediction modules. For this sequential process, Charnes, Cooper, and Rhodes and Banker, Chames and Cooper DEA models and back propagation neural network (BPNN) are used. The proposed methodology is empirically supported using longitudinal data of Japanese electronics manufacturing firms. Findings – The combined modeling approach proves effective through sequential processes by streamlining DEA analysis and BPNN predictions. The DEA model captures notable characteristics and efficiency trends of the Japanese electronics manufacturing industry and extends its utility as a preprocessor to neural network prediction modules. BPNN, in conjunction with DEA, demonstrates promising estimation capability in predicting efficiency scores and best performance benchmarks for DMUs under evaluation. Research limitations/implications – Integration of adaptive prediction capacity into the measurement model is a practical necessity in the benchmarking arena. The proposed framework has the potential to recalibrate benchmarks for firms through longitudinal data analysis. Originality/value – This research paper proposes an innovative approach of performance measurement and prediction in line with superiority-driven best performance modeling. Adaptive prediction capabilities embedded in the proposed model enhances managerial flexibilities in setting performance goals and monitoring progress during pursuit of improvement initiatives. This paper fills the research void through methodological breakthrough and the resulting model can serve as an adaptive decision support system. [ABSTRACT FROM AUTHOR]

The recent literature indicates that face recognition and facial expression classification has achieved a high accuracy on benchmark datasets with a large number of face images in the wild. On the other hand, unlike the purpose of recognizing as many people as possible, real applications for families or companies usually aim to recognize a small group of people as accurate as possible. In case of the face is partially occluded, convolutional solutions always simply put images with occlusions into the training dataset and hope the convolution neural network learn a model robust to partial occlusion. These processes not only increase the burden of learning, but also affect the model to identify normal images without occlusions. In this paper, we investigate the research problem of facial image reconstruction to discard the influence of partial occlusion. Based on the phenomenon that human faces are roughly symmetrical, we propose to reconstruct the facial information of occluded areas with the intact half face. Specifically, occlusion on the left‐half face is reconstructed with a linear combination of features on the right‐half face and vice versa. The process is modeled by keeping row sparsity for the coefficient matrix with l2,1‐norm regularization while minimizing the reconstruction error. An alternative iterative algorithm is proposed to solve the optimization problem. To validate the effectiveness of the reconstruction, we fine‐tune a pretrained model, AlexNet, training on normal face images and test with various occluded images. Extensive experimental results show that our method has improved the classification performance effectively. © 2021 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

The conventional artificial neural network (ANN) model is an effective way to detect harmonic signals. However, the solution of the model cannot converge when the initial value of the excitation function (IVEF) is not suitable, causing a large error in harmonic detection, especially in the case of power system noise. We improve the back propagation (BP) neural network (BNN) model by using an adjustable excitation function and the basic inertial algorithm in this paper. Based on this model and combined with the glowworm swarm optimization (GSO) algorithm, a harmonic/inter‐harmonic detection method is proposed. The network first uses an improved GSO algorithm with an adaptive step to optimize the IVEF of BNN. Then, BNN is trained at this initial value, so that amplitude, phase, and frequency of harmonics/inter‐harmonics can be obtained. The simulation data analyses show that that the method has good stability, convergence, strong anti‐noise ability, and ultrahigh detection accuracy. It can accurately separate the integer and non‐integer harmonics. Compared with the traditional ANN algorithm and Hanning‐FFT, the detection accuracy of this algorithm can be improved by 2–4 orders of magnitude. © 2020 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

As a criterion of a good firm, a lucrative and growing business has been said to be important. Recently, however, high profitability and high growth potential are insufficient for the criteria, because social influences exerted by recent firms have been extremely significant. In this paper, high social relationship is added to the list of the criteria. Empirical corporate social performance versus corporate financial performance (CSP–CFP) relationship studies that consider social relationship are very limited in Japan, and there are no definite conclusions for the studies in the world, because of scant data and the inappropriate methods, especially for supporting linear hypothesis which these studies are based on. In this paper, the CSP–CFP relationship is analyzed by an artificial neural networks model, which can deal with a non-linear relationship, using 10-year follow-up survey data. [ABSTRACT FROM AUTHOR]

In this research, it was tried to study the effect of economic, social and political development of the country in improving the quality of financial reporting of companies by developing a comprehensive and scientific model by setting Japan as a model, emphasizing the efficiency and effectiveness of the technique--management accounting as a mediating variable. The mentioned model was developed through structural equation modelling with PLS software. This research aims to collect economic, social and political data from the data of international reports and to collect data related to the quality of the internal information environment (alternative index to measure the efficiency and effectiveness of management accounting techniques and tools) and The quality of financial reporting has been used from the financial statements of all companies in the automotive industry, parts manufacturing, electronic equipment and heavy vehicles in Tehran and Tokyo Stock Exchanges. The developed model and the results of the research hypothesis test showed that the main factors of development, including economic, social and political factors, have a significant positive effect on improving the quality of the internal information environment through the efficiency and effectiveness of management accounting techniques and tools, while the factors The main development, neither directly nor indirectly through the mediating variable of the quality of the internal information environment, has a significant effect on the quality of financial reporting of companies. The same result was repeated in the fourth hypothesis regarding the country of Japan, and it showed that economic, social and political factors in both Iran and Japan cannot lead to improving the quality of financial reporting. [ABSTRACT FROM AUTHOR]

In this paper, we propose a neural network approach for next day load curve forecasting based on similarity. The proposed method has the advantage of dealing with not only the nonlinear part of load curve but also with weekend and special day features. The proposed neural network is used to modify the load curve of a similar day by using temperature information. The suitability of the proposed approach is illustrated through an application to actual load data of Okinawa Electric Power Company in Japan. [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]

This paper provides a cluster analysis of the bank lending behavior in Japan from 1977 to 2002, in which the bubble economy occurred. The self-organizing map is used for the clustering. This clustering method is a type of artificial neural network and enables us to visualize the similarities among the high-dimensional data on bank lending. A result of clustering indicates that this time period is classified into two periods: before and after 1992. Moreover, we successfully extract the characteristics of bank lending behavior on 8 borrower sectors from another clustering result. [ABSTRACT FROM AUTHOR]

This paper proposes a procedure to estimate tsunami wave forces on coastal bridges through a novel method based on Extreme Learning Machine (ELM) and laboratory experiments. This research included three water depths, ten wave heights, and four bridge models with a variety of girders providing a total of 120 cases. The research was designed and adapted to estimate tsunami bore forces including horizontal force, vertical uplift and overturning moment on a coastal bridge. The experiments were carried out on 1:40 scaled concrete bridge models in a wave flume with dimensions of 24 m x 1.5 m x 2 m. Two six-axis load cells and four pressure sensors were installed to the base plate to measure forces. In the numerical procedure, estimation and prediction results of the ELM model were compared with Genetic Programming (GP) and Artificial Neural Networks (ANNs) models. The experimental results showed an improvement in predictive accuracy, and capability of generalization could be achieved by the ELM approach in comparison with GP and ANN. Moreover, results indicated that the ELM models developed could be used with confidence for further work on formulating novel model predictive strategy for tsunami bore forces on a coastal bridge. The experimental results indicated that the new algorithm could produce good generalization performance in most cases and could learn thousands of times faster than conventional popular learning algorithms. Therefore, it can be conclusively obtained that utilization of ELM is certainly developing as an alternative approach to estimate the tsunami bore forces on a coastal bridge. [ABSTRACT FROM AUTHOR]

This paper presents a method for designing neural networks using a genetic algorithm (GA) with deterministic mutation (DM) based on learning. The GA presented in this paper has a large framework including DM, which is performed on the basis of the results from neural network learning. It can achieve better convergence properties than traditional GAs. This framework is an evolutional system based on mutual interaction between DM and traditional genetic operators including stochastic mutation. It is also a model of transcription and reverse transcription in DNA. We show that the present method is better than conventional GAs with respect to convergence in learning. © 1998 Scripta Technica. Syst. Comp Jpn, 29(3): 10–17, 1998 [ABSTRACT FROM AUTHOR]

Most of the neural network models based on the Lyapunov stability contain various problems such as the trap of the local minimum, and the limit of their dynamic performances has been pointed out. This paper attempts to provide a new dynamic performance to such a neural network model by introducing chaos dynamics. The features of the chaotic dynamic model proposed here is that the dynamical equation describing the trajectory on the energy curve has a periodically varying nonlinear resistance in the dissipation term. By iterating the stable and unstable phases, the chaotic transitions of the state can be realized. The proposed dynamic model is applied to the error backpropagation learning and the memory recall in the Hopfield-type network, and the chaotic minimum transitions in the dynamic process are verified, it is verified further that the intermittent chaos generated by an appropriate parameter manipulation can introduce useful dynamic behaviors into the network, e.g., the flexible learning and the memory recall with a structure containing both stability and plasticity. [ABSTRACT FROM AUTHOR]

Traffic calming devices (TCD) like speed bumps are effective measures in reducing speed on residential roads but in a limited section of the road. Previous studies found that if the entrance speed of the traffic calming road (TCR) is higher than the posted speed limit (30 kmph), it will affect speed reduction by TCD. Little research has been undertaken on predicting speed at the entrance of a TCR. Thus, the purpose of this study is to predict the variables of interest in the context of the study area by applying a methodological approach described in this study; a combination of multiple linear regression, and artificial neural networking (ANN) models. Seventeen variables derived from four categories including road design, roadside environment, traffic, and speed were considered in the proposed model. Two approaching criterion was defined in this study: a minor (speed limit 30 km/h) and a major approach (speed limit 40/50 km/h). Over 1000 vehicle speeds were collected from almost 20 different urban roads in Japan. Entrance characteristics, lane width, barrier types, posted speed limit, etc. have found as significant predictors in the regression output. Furthermore, the ANN technique was employed where the input variables were the significant predictors of regression models, and variable importance was also calculated using sensitivity analysis. Although the results of sensitivity and regression are different, sensitivity plays a role in illuminating the relationship between road factors and speed. By comparing the prediction performance between the two models this study reveals that developing an ANN model followed by regression output could be a consistent method for analyzing the entry speed of TCR for varying approaches. Additionally, the findings can be used to effectively predict the entry speed of TCR without installing a gateway design and help to achieve a safer speed through modifications to the road features. [ABSTRACT FROM AUTHOR]

Recently, the fine‐grained geolocalization of User‐Generated Short Text (UGST) has been increasingly attracting much attention. Accurate geolocation can benefit many applications, especially for the location‐based services. However, since the majority of UGSTs are short, noisy and not geotagged, existing methods still suffer from two issues. One is the heavy reliance on the manual features not fully exploiting the semantic information. Another is the free writing style of social media resulting in extremely few useful geo‐indicative information. To address these issues, we propose a novel Fine‐grained Geolocalization method for UGSTs with Preprocessing, location‐entity consistency Replacing, Filtering, Convolutional neural network (FG‐PRFC), which only relies on UGST itself. Compared to existing methods, FG‐PRFC has four unique characteristics: (1) We present a UGST‐oriented preprocessing method to obtain more semantic information. (2) To tackle the abbreviation issue, we develop a replacing method to allow geo‐indicative words behaving in the same way. (3) Following the idea of TFIDF, we weight the words in UGST and then develop a location‐free UGST filtering method. (4) We employ convolutional neural network to model the relationship between words and locations. Extensive experiments on three ground‐truth datasets demonstrate that our method has a significant improvement compared to state‐of‐art methods. © 2022 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

The evaluation of the representation of the surface downward shortwave flux (DSF) from atmospheric reanalysis data products is required to obtain reliable information for the resource assessment of surface solar energy. The representation of the DSF from the Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2), reanalysis data product was evaluated using surface solar radiation from ground-based observations in Japan. The cloud fraction (CFR) and cloud optical thickness (COT) from Moderate Resolution Imaging Spectroradiometer (MODIS) were also used as references. The CFR from MERRA-2 tends to be smaller than that from MODIS, and the correlation between the difference in the CFR and that in the DSF is negative. The correlation between the difference in the COT and that in the DSF is weakly negative. To quantify the effects of the difference in the CFR and COT to that of the DSF, a regression model based on an artificial neural network architecture that emulates the process of the DSF in MERRA-2 was constructed. Numerical experiments using the emulator quantify contributions of each of the differences in the CFR and COT and joint contributions of the two variables. In addition, a cluster analysis was performed to clarify the differences in the seasonal changes in the monthly mean bias error (MBE) in the DSF among ground observation stations, and three clusters were identified. Contributions of the differences in the CFR and COT to the seasonal change in the monthly MBE were also clarified using the results of the numerical experiments. [ABSTRACT FROM AUTHOR]

Aim: The evaluation of the effects of resuscitation activity factors on the outcome of out-of-hospital cardiopulmonary arrest (OHCA) requires consideration of the interactions among these factors. To improve OHCA success rates, this study assessed the prognostic interactions resulting from simultaneously modifying two prehospital factors using a trained machine learning model. Methods: We enrolled 8274 OHCA patients resuscitated by emergency medical services (EMS) in Nara prefecture, Japan, with a unified activity protocol between January 2010 and December 2018; patients younger than 18 and those with noncardiogenic cardiopulmonary arrest were excluded. Next, a three-layer neural network model was constructed to predict the cerebral performance category score of 1 or 2 at one month based on 24 features of prehospital EMS activity. Using this model, we evaluated the prognostic impact of continuously and simultaneously varying the transport time and the defibrillation or drug-administration time in the test data based on heatmaps. Results: The average class sensitivity of the prognostic model was more than 0.86, with a full area under the receiver operating characteristics curve of 0.94 (95% confidence interval of 0.92–0.96). By adjusting the two time factors simultaneously, a nonlinear interaction was obtained between the two adjustments, instead of a linear prediction of the outcome. Conclusion: Modifications to the parameters using a machine-learning-based prognostic model indicated an interaction among the prognostic factors. These findings could be used to evaluate which factors should be prioritized to reduce time in the trained region of machine learning in order to improve EMS activities. [ABSTRACT FROM AUTHOR]

Abstract: The “fluid-flow tomography”, an advanced technique for geoelectrical survey based on the conventional mise-à-la-masse measurement, has been developed by Exploration Geophysics Laboratory at the Kyushu University. This technique is proposed to monitor fluid-flow behavior during water injection and production in a geothermal field. However data processing of this technique is very costly. In this light, this paper will discuss the solution to cost reduction by applying a neural network in the data processing. A case study in the Takigami geothermal field in Japan will be used to illustrate this. The achieved neural network in this case study is three-layered and feed-forward. The most successful learning algorithm in this network is the Resilient Propagation (RPROP). Consequently, the study advances the pragmatism of the “fluid-flow tomography” technique which can be widely used for geothermal fields. Accuracy of the solution is then verified by using root mean square (RMS) misfit error as an indicator. [Copyright &y& Elsevier]

This study examines the export performance of 350 randomly selected Japanese engineering service firms. Inquiry is made regarding environmental, organizational, managerial, strategic, and functional determinants of these exporting firms. Two primary questions are addressed. Can performance category membership of engineering service exporters be predicted? Second and foremost, which determinants are the most effective in differentiating between the export performance categories? An Artificial Neural Network is selected as the statistical method because of the different perspective it provides for a highly non-linear function having many variables, offering results that consistently prove to numerically approximate such functions much easier than conventional methods, together with the ability to dependably and accurately predict membership classification while providing weighted analyses of input variables. Findings suggest that each group's membership can be consistently and accurately predicted and further identify which determinants dominantly impact category membership as supported by differences in the feature extraction phase of the neural network approach. [ABSTRACT FROM AUTHOR]

MRS (magnetic resonance spectroscopy) consists of collecting the nuclear magnetic resonance (NMR) spectrum of the metabolites in the living body and estimating it. In the field of MRS, we usually get an NMR spectrum by applying the Fourier transform to the real part of the NMR signal [free induction decay (FID) signal], which is a complex-valued signal. We must then measure the area of the spectral peak to estimate metabolites in the living body from the spectrum. A curve fitting technique is used for that measurement. However, this technique is not suitable for large quantities of data processing, because human intervention is necessary and is laborious. For this reason, an automatic method of estimating an NMR spectrum is required in the field of MRS. In this paper, we propose a method of estimation of an NMR spectrum by using a Hopfield complex-valued neural network and try to estimate the spectrum automatically. In addition, we consider damping state of the NMR signal and devise a method called sequential extension of section (SES). It is shown that the estimation precision of the spectrum improves when SES is used. Furthermore, SES is found to reduce the local minimum problem in Hopfield neural networks. © 2004 Wiley Periodicals, Inc. Syst Comp Jpn, 35(10): 14–22, 2004; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/scj.10705 [ABSTRACT FROM AUTHOR]

This paper proposes a new neural network (the concepts-of-sentences associative neural network, CSANN) which performs memory and recall of sentences represented by a semantic network, as an approach toward natural language processing. The semantic network is represented in terms of the triples representation of concepts. The triples representation of concepts consists of concepts linked by relations in sentences represented by natural language. CSANN handles sentence data, which are difficult to handle since the data size is not unique, by decomposing the data into triples representations of concepts and integrating them into the semantic network format. It is verified by computer simulation that the proposed network can store sentences and recall sentences even if the input information is incomplete. Thus, the effectiveness of the proposed network in natural language processing is demonstrated. © 2003 Wiley Periodicals, Inc. Syst Comp Jpn, 34(11): 43–54, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/scj.10476 [ABSTRACT FROM AUTHOR]

This paper reports simple numerical experiments of the application of multi-layered and feed-forward neural networks to regression with change points to clarify one of the effectiveness of the neural network model compared with non-parametric regression methods based on scatter plot smoothing. We also show an illustrative example, which successfully draws too rapid growth of GDP in Japan at the bubble economy around 1990 by interpreting decomposition of regression function suggested by the optimal neural networks fitting. Copyright © 2002 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

The machine learning technique, namely artificial neural networks (ANN), is used to predict the surface air temperature (SAT) anomalies over Japan in the winter months of December, January, and February for the period 1949/50–2019/20. The predictions are made for the four regions Hokkaido, North, Central, and West of Japan. The inputs to the ANN model are derived from the anomaly correlation coefficients among the SAT anomalies over the regions of Japan and the global SAT and sea surface temperature anomalies. The results are validated using anomaly correlation coefficient (ACC) skill scores with the observation. It is found that the ANN predictions over Hokkaido have higher ACC skill scores compared to the ACC scores over the other three regions. The ANN-predicted SAT anomalies are compared with that of ensemble mean of eight of the North American Multimodel Ensemble (NMME) models besides comparing them with the persistent anomalies. The ANN predictions over all the four regions have higher ACC skill scores compared to the NMME model skill scores in the common period of 1982/83–2018/19. The ANN-predicted SAT anomalies also have higher hit rate and lower false alarm rate compared to the NMME-predicted SAT anomalies. All these indicate that the ANN model is a promising tool for predicting the winter SAT anomalies over Japan. [ABSTRACT FROM AUTHOR]

We propose a new algorithm, focusing on spatial amplitude patterns, to automatically detect volcano seismic events from continuous waveforms. Candidate seismic events are detected based on signal-to-noise ratios. The algorithm then utilizes supervised machine learning to classify the existing candidate events into true and false categories. The input learning data are the ratios of the number of time samples with amplitudes greater than the background noise level at 1 s intervals (large amplitude ratios) given at every station site, and a manual classification table in which 'true' or 'false' flags are assigned to candidate events. A two-step approach is implemented in our procedure. First, using the large amplitude ratios at all stations, a neural network model representing a continuous spatial distribution of large amplitude probabilities is investigated at 1 s intervals. Second, several features are extracted from these spatial distributions, and a relation between the features and classification to true and false events is learned by a support vector machine. This two-step approach is essential to account for temporal loss of data, or station installation, movement, or removal. We evaluated the algorithm using data from Mt. Ontake, Japan, during the first ten days of a dense observation trial in the summit region (2017 November 1–10). Results showed a classification accuracy of more than 97 per cent. [ABSTRACT FROM AUTHOR]

This paper proposes a quasi-competitive network in which the mass reaction intensity of the elements in the intermediate layer is kept constant as a model that approximates the continuous nonlinear function. As the loss function, the local loss function is considered. It is composed of the error and the local model loss term reflecting the local model size. The learning algorithm of the quasi-competitive network including the change of the number of elements is derived from the local loss function. It is shown by numerical experiment that the quasi-competitive network can form quickly the structure reflecting the target function and has a high generalization power. It is shown also that the high generalization power is due to the constant mass reaction intensity of the elements in the intermediate layer of the quasi-competitive network. [ABSTRACT FROM AUTHOR]

Although it is pointed often that multilayer neural networks should have a certain degree of fault tolerance, very few discussions based on the rigorous definition of fault tolerance have been made so far. Also, there have been few discussions on the mechanisms that bring out the fault tolerance. This paper shows that a learning algorithm that directly reduces a measure of fault tolerance can be derived in a similar way to the conventional backpropagation. By analyzing the resulting networks, the mechanism that realizes the fault tolerance and the properties of the fault-tolerant networks are investigated. Simulation results show the effectiveness of the proposed learning algorithm. It also is revealed that the utilization of the redundant hidden units and the saturation property of the sigmoid function realizes the fault tolerance. Moreover, it is shown that a good influence on generalization ability can be expected from the learning algorithm for fault tolerance. [ABSTRACT FROM AUTHOR]

Self-compacting concrete originated from Japan in the late 80's. because it does not require the vibrator to be consolidated. it was mostly used from 2000.Since there is no specific code for the mix design. The application of ANN will help to predict exactly the mix design of SCC. The application of artificial neural networks in self-compacting concrete began from the year of 2001.From the beginning the study shows that the ANN is a reliable method to predict the optimum admixture in a fast manner. When compared to do experimentally which is costlier and time consuming. ANN is programmed with the two modulus that is the output of the first ANN program will be the input of the second program are mostly used. Self-compacting concrete has been produced using the different types of mineral and chemical admixtures in order to reduce the material wastage and time. Glass powder is an excellent admixture. Using glass powder will increase the strength of SCC by 15 % and it also reduces the absorption of carbon dioxide. There should be application of ANN in SCC made with admixture of glass powder will help to develop an eco-friendly concrete. [ABSTRACT FROM AUTHOR]

Utilizing the self‐organizing map (SOM), a type of artificial neural network, a new classification of the climate of Japan is proposed. The SOM is applied on the monthly mean surface air temperature (SAT) anomalies, extracted from 762 stations. Considering the strong seasonal cycle in mid‐latitudes, the classification is performed for two distinct seasons, boreal winter and boreal summer. Applied on monthly average temperature, to capture the seasonal signal, the SOM is an easily implementable interesting tool to (a) objectively capture the patterns present in the input data, and (b) identify the source of interannual variability, which is crucial for power demand forecasting. While modulated by local conditions, SAT in Japan is mainly influenced by large‐scale circulation. It is found in this study that stronger relationships exist for tropics with southern regions and for extra‐tropics with northern regions, in the seasonally oriented teleconnection patterns. In winter, the regions are organized along a north–south orientation, with a secondary west–east orientation in the central part of the country. This organization is a function of the strength of the link with the tropical Pacific and Indian Oceans sea‐surface temperatures. Changes in the speed of the westerly jet, together with the modulation of the Western Pacific (WP)‐like and Eurasian (EU)‐like patterns, are also important for SAT in Japan in winter. In summer, the main organization follows a west–east orientation. The Pacific Japan (PJ)‐like pattern plays an important role in the geographical distribution of the SAT, together with the existence of mid‐latitude and subtropical wave trains, like the Silk Road (SR)‐like pattern. [ABSTRACT FROM AUTHOR]

Probabilistic solar irradiance forecasting has received widespread attention in recent years, as it provides more uncertainty information for the future photovoltaic generation. In this study, a hybrid probabilistic solar irradiance prediction method is proposed, which combines a deep recurrent neural network and residual modeling. Specifically, the long short‐term memory‐based point prediction using historical records and related features is applied to obtain deterministic forecasts. Next, these deterministic forecasts are employed as inputs to estimate the residual distributions. Furthermore, maximum likelihood estimation is utilized to compute the parameters of the residual distribution. Finally, the point prediction and residual distribution jointly generate the final probabilistic forecasting results. Compared with other deterministic and probabilistic forecasting models, the proposed method yields promising results on a publicly available dataset. © 2020 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC. [ABSTRACT FROM AUTHOR]

The time difference between coordinated universal time (UTC) and a hydrogen maser, which is a master oscillator for the local realization of UTC at the National Metrology Institute of Japan (NMIJ), has been predicted by using one of the deep learning techniques called a one-dimensional convolutional neural network (1D-CNN). Regarding the prediction result obtained by the 1D-CNN, we have observed improvement in the accuracy of prediction compared with that obtained by the Kalman filter. Although more investigations are required to conclude that the 1D-CNN can work as a good predictor, the present results suggest that the computational approach based on the deep learning technique may become a versatile method for improving the synchronous accuracy of UTC(NMIJ) relative to UTC. [ABSTRACT FROM AUTHOR]

A method was developed to derive an NARX model from a neural network so that the usability of open‐source libraries for network learning was combined with the NARX advantage of revealing the system structure. After the neural network model was trained on input and output data, the sigmoid activation functions were expanded into Taylor series. Candidate parameters in the NARX model were calculated from the connection weights in the neural network and coefficients in the series. The NARX model structure was determined by the extended least‐squares (ELS) method and the Bayesian information criterion (BIC). Correct NARX models were successfully detected in computer simulations and an experiment. The developed method can be used for any activation functions, including the sigmoid function, if they are expanded into Taylor series. © 2019 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc. [ABSTRACT FROM AUTHOR]

Several drowsiness detection technologies have been developed to combat traffic accidents. Drowsiness evaluation has been attempted using the time‐series changes in nasal skin temperature. However, constructing a detection model based on this has been difficult because recent studies have reported an individual difference in skin temperature behaviors. In this study, the conventional method of extracting features was revised. The model of the level of drowsiness was constructed based on facial skin temperature distribution using a convolutional neural network (CNN). The drowsiness level was developed by the New Energy and Industrial Technology Development Organization as an objective drowsiness evaluation index based on facial expressions. With CNNs, features related to learning can be observed as feature quantity distributions. As a result, a general model created has a lack of generality and it is thought that not only the response to drowsiness, but also face shape, exhibit individual differences. Consequently, different features were found in each subject. Through feature maps in individual models, it is believed that skin temperature changes have both reproducible and individual response characteristics to drowsiness, due to the asymmetric left–right change in feature quantity distribution depending on the observed drowsiness level. This method was compared with the conventional method of extracting other features related to drowsiness. It suggested that the skin temperature of not only the nasal region but also the entire face changes as drowsiness increases. Consequently, each discrimination rate calculated by the CNN was at least 20% higher than that obtained via conventional methods. © 2019 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc. [ABSTRACT FROM AUTHOR]

Along with the rapid increase in wind power penetration into the power grid, wind power generation predicting is becoming increasingly important to power system operators and electricity market participants. However, the random nature of the wind power would increase the uncertainty of power systems. The influencing factor is one of the most important factors in the quality of wind power prediction. In order to obtain a higher prediction accuracy, a two‐stage prediction method combined with meteorological factor and fault time is proposed. In the first stage, we present a detailed review of fault time of wind power predicting with machine‐learning methods and compare ten different models based on different influencing factors, and then, the second stage is combined with the predicted results of first stage, coupled with meteorological factors for the final wind power prediction. The results show good prediction accuracy in operation data of one wind farm from the Hu‐Nan province with different machine‐learning algorithms. © 2019 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc. [ABSTRACT FROM AUTHOR]

After the severe nuclear disaster in Fukushima, which was triggered by the Great East Japan earthquake in March 2011, nuclear power plants in Japan were temporarily shut down for mandatory inspections. To prevent large-scale blackouts, the Japanese government requested companies and households to reduce electricity consumption in summer and winter. It is reported that the domestic electricity demand had a structural decrease because of the electricity conservation effect (ECE). However, quantitative analysis of the ECE is not sufficient, and especially time variation of the ECE remains unclear. Understanding the ECE is important because Japan’s NDC (nationally determined contribution) assumes the reduction of CO2 emissions through aggressive energy conservation. In this study, we develop a time series model of monthly electricity demand in Japan and estimate time variation of the ECE. Moreover, we evaluate the impact of electricity conservation on CO2 emissions from power plants. The dynamic linear model is used to separate the ECE from the effects of other irrelevant factors (e.g. air temperature, economic production, and electricity price). Our result clearly shows that consumers’ electricity conservation behavior after the earthquake was not temporary but became established as a habit. Between March 2011 and March 2016, the ECE on industrial electricity demand ranged from 3.9% to 5.4%, and the ECE on residential electricity demand ranged from 1.6% to 7.6%. The ECE on the total electricity demand was estimated at 3.2%–6.0%. We found a seasonal pattern that the residential ECE in summer is higher than that in winter. The emissions increase from the shutdown of nuclear power plants was mitigated by electricity conservation. The emissions reduction effect was estimated at 0.82 MtCO2–2.26 MtCO2 (−4.5% on average compared to the zero-ECE case). The time-varying ECE is necessary for predicting Japan’s electricity demand and CO2 emissions after the earthquake. [ABSTRACT FROM AUTHOR]

The increasing frequency of devastating floods from heavy rainfall—associated with climate change—has made river stage prediction more important. For steep, forest-covered mountainous watersheds, deep-learning models may improve prediction of river stages from rainfall. Here we use the framework of multilayer perceptron (MLP) neural networks to develop such a river stage model. The MLP is constructed for the Shimanto river, which lies in southwestern Japan under a mild, rain-heavy climate. Our input for stage estimation, as well as prediction, is a long-term rainfall time series. With a one-year time series of rainfall, the model estimates the stage with RMSE less than 67 cm for about 10 m of stage peaks, as well as accurately simulating stage-time fluctuations. Furthermore, the forecast model can predict the stage without rainfall forecasts up to three hours ahead. To estimate the base flow stages as well as flood peaks with high precision, we found that the rainfall time series should be at least one year. This indicates that the use of a long rainfall time series enables one to model the contributions of ground water and evaporation. Given that the delay between the arrival time of rainfall at a rain-gauge to the outlet change is well-simulated, the physical concepts of runoff appear to be soundly embedded in the MLP. [ABSTRACT FROM AUTHOR]

Background: Patients with acute liver failure are classified according to the interval between the onset of hepatitis symptoms and the development of hepatic encephalopathy. We examined the validity of such classifications. Methods: The subjects were 1,022 patients enrolled in a nationwide survey in Japan. The intervals between the onset of the hepatitis symptoms and the development of encephalopathy were 10 days or less in 472 patients (group-A), between 11 and 56 days in 468 patients (group-B), and longer than 56 days in 82 patients (group-C). Data on a total of 104 items collected from the patients were subjected to clustering using a self-organizing map. Results: The patients were classified into three clusters. The first cluster consisted of 411 patients (group-A: 57%, group-B: 39%, group-C: 4%). Their incidence of complications was low; 34% underwent liver transplantation (LT), and their survival rate was 90%, while 94% of those treated without transplant were rescued. The second cluster consisted of 320 patients (21, 65, and 14% groups A, B, and C, respectively), who showed a high incidence of complications; the survival rate was 7% in the patients treated conservatively without LT. Sixteen percent underwent LT and survival rate of these patients was 52%. There was a third cluster, of 291 patients (59, 34, and 7% groups A, B, and C, respectively). Without LT, 81% of the patients died. Seven percent were treated by LT and their survival rate was 60%. Conclusions: Clustering revealed that patients with acute liver failure could be classified into three clusters independent of the interval between the onset of disease symptoms and the development of encephalopathy. This technique may be useful, since the outcomes of the patients differed markedly among the clusters. [ABSTRACT FROM AUTHOR]

This study examined the validity of using an artificial neural network (ANN) to predict tsunami water levels at several locations in Osaka Bay. The metropolitan areas of Osaka Bay have short warning times for tsunamis; a real-time tsunami forecast will allow for improved evacuation plans and will reduce the effect of these coastal disasters. Different tsunami conditions changing the relative strength of the asperities and background sources, such as fault displacement, fault length, fault width, fault slope, depth from sea bottom, and strike, were used for training the ANN; the data sets were generated by applying the nonlinear shallow water wave equations assuming different earthquake fault models. The linear activation function produced optimal results for the ANN output units, and the tangent-sigmoid function yielded good results for the ANN hidden layer units. The Levenberg-Marquardt method with Bayesian regulation was employed for the training of the ANN. Output from the trained ANN was the preliminary and secondary tsunami waves; these ANN output data agreed well with numerically obtained tsunami simulation results. [ABSTRACT FROM AUTHOR]

Prediction of pore pressure change is an effective tool to properly monitor changes of groundwater flow caused by any construction work in fractured rock mass. Due to the complexity of hydrogeologic conditions in fractured rock and the scale of interest of the study domain, prediction of pore pressure changes by numerical models has not been precise enough to meet monitoring requirements. Considering these problems, a Grey model that combines the finite element method (FEM) and the artificial neural network (ANN) was developed for more precise prediction of pore pressure changes. In this model, several patterns of pore pressure changes were calculated by FEM for a simplified hydrogeologic conceptual model at a scale smaller than a representative elementary volume. The ANN model was then constructed to predict the actual pore pressure change using these FEM results as inputs. This modeling approach was adopted to predict the pore pressure changes caused by the construction of shafts of Mizunami Underground Research Laboratory (MIU), Japan. From the results obtained for MIU, it can be concluded that the proposed Grey model is a powerful tool for monitoring of pore pressure changes. [ABSTRACT FROM AUTHOR]

In order to minimize time and capital investment for preliminary designs of active landslides, researchers have proposed correlation charts relating soil index properties such as LL, PI and CF to residual friction angle. However, estimated residual friction angle shows large variations by chart type and soil properties. This is because the databases that have been used to establish these correlation charts belongs to different soil types and obtained using different testing apparatus with different shearing rates. Therefore, they cannot be generalized. In this study, an artificial neural network (ANN) model was developed to evaluate databases for soils from Britain, China, Japan and the Pacific Islands to determine which soil index properties give best estimate of residual friction angle of these soils. The ANN model developed was further validated using Skempton's (1985) data. The results indicate that there exist a reasonable correlation between soil index properties such as LL, PI, and CF and residual friction, providing that the soils are tested in similar manner and have similar mineralogy. The sensitivity analysis results indicated that residual friction angle is most dependent on CF of soils. The results also show that the ANN model developed is a powerful for predicting residual friction angle of soils using soil index properties. [ABSTRACT FROM AUTHOR]