Search

This study aims to generate a deep learning algorithm-based model for quantitative prediction of financial losses due to accidents occurring at apartment construction sites. Recently, the construction of apartment buildings is rapidly increasing to solve housing shortage caused by increasing urban density. However, high-rise and large-scale construction projects are increasing the frequency and severity of accidents occurring inside and outside of construction sites, leading to increases of financial losses. In particular, the increase in severe weather and the surge in abnormal weather events due to climate change are aggravating the risk of financial losses associated with accidents occurring at construction sites. Therefore, for sustainable and efficient management of construction projects, a loss prediction model that prevents and reduces the risk of financial loss is essential. This study collected and analyzed insurance claim payout data from a main insurance company in South Korea regarding accidents occurring inside and outside of construction sites. Deep learning algorithms were applied to develop predictive models reflecting scientific and recent technologies. Results and framework of this study provide critical guidance on financial loss management necessary for sustainable and efficacious construction project management. They can be used as a reference for various other construction project management studies., (© 2022. The Author(s).)

Signal and noise classification can add an extra level of constraint for earthquake phase picking by pinpointing the signal waveforms from continuous seismic data for more accurate arrival picking. However, the continuously increasing data collected by worldwide stations exceeds the ability of manual analysis. Moreover, manual earthquake data analysis depends on seismologists' expert knowledge, resulting in inconsistent analysis results. To address this, we proposed a generalized deep learning (DL) network architecture to discriminate earthquake signal and noise waveforms. The proposed DL framework is a novel architecture comprising a feature extractor, a classifier and two hybrid attention modules. It utilizes different kernel sizes for more detailed feature extraction, and the hybrid attention mechanism module can guide the network to focus more on the waveform characteristics. To illustrate the power of the proposed DL network, we applied it to classify the earthquake signal and noise of the 3-C Texas Earthquake Dataset. The results demonstrate that the accuracy of the proposed method in the testing set reaches 99.83 per cent. We further utilize the transfer learning strategy to demonstrate the transferability of the proposed network with the Stanford earthquake data set, showing an encouraging classification accuracy of 95.03 per cent. Additionally, we conducted an additional experiment on arrival picking by integrating decoder blocks into the classification network, which achieves remarkable P - and S -wave arrival picking accuracy. [ABSTRACT FROM AUTHOR]

This study presents a new tool for predicting the compressive strength of alkali-activated concrete (AAC) based on its binder mineralogy. It was made using a machine-learning (ML) framework. To achieve this challenging task, the authors collected 45 data sources from the literature to build a data set of 809 samples that included nine effective features, such as binder content, alkaline-to-binder ratio, binder chemical compositions (CaO, SiO2 , Al2O3 , and MgO contents), NaOH molarity and percentage in alkaline solution, age, and compressive strength. To assess the accuracy of the prediction tool, the authors trained and evaluated the data set using the most relevant ML methods: Lasso regression, random forest regression, decision tree, AdaBoost, extreme gradient boosting (XGB), and long short-term memory with recurrent neural network (LSTM-RNN). An experimental program was also conducted using the ML approaches to further validate the accuracy of the predictions. Overall, the XGB and LSTM-RNN methods were observed to significantly outperform the other methods in terms of accuracy when predicting compressive strength. Particularly impressive results were seen, with R2 values of 0.93 and 0.96 for compressive strength prediction being recorded. Further analysis of the binder mineralogy showed that increasing calcite content percentage led to an increase in AAC compressive strength, whereas increasing silicates in the binder mineralogy caused a decrease in AAC compressive strength due to the shortage of calcites. The Shapley additive explanations (SHAP) analysis revealed that calcite and silicate had the highest SHAP values for the AAC compressive strength. In contrast, the Al2O3 and MgO percentages had only a minor impact on the compressive strength of AAC. [ABSTRACT FROM AUTHOR]

Copyright of Progress in Earthquake Sciences is the property of China Earthquake Administration, Institute of Geophysics and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Earthquakes are classified as one of the most devastating natural disasters that can have catastrophic effects on the environment, lives, and properties. There has been an increasing interest in the prediction of earthquakes and in gaining a comprehensive understanding of the mechanisms that underlie their generation, yet earthquakes are the least predictable natural disaster. Satellite data, global positioning system, interferometry synthetic aperture radar (InSAR), and seismometers such as microelectromechanical system, seismometers, ocean bottom seismometers, and distributed acoustic sensing systems have all been used to predict earthquakes with a high degree of success. Despite advances in seismic wave recording, storage, and analysis, earthquake time, location, and magnitude prediction remain difficult. On the other hand, new developments in artificial intelligence (AI) and the Internet of Things (IoT) have shown promising potential to deliver more insights and predictions. Thus, this article reviewed the use of AI-driven Models and IoT-based technologies for the prediction of earthquakes, the limitations of current approaches, and open research issues. The review discusses earthquake prediction setbacks due to insufficient data, inconsistencies, diversity of earthquake precursor signals, and the earth's geophysical composition. Finally, this study examines potential approaches or solutions that scientists can employ to address the challenges they face in earthquake prediction. The analysis is based on the successful application of AI and IoT in other fields. [ABSTRACT FROM AUTHOR]

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

Obtaining reliable low‐frequency seismic data is crucial for effectively reducing cycle‐skipping in full waveform inversion. However, acquiring high signal‐to‐noise ratio low‐frequency information from field data remains a challenge. An effective solution to mitigate cycle‐skipping is to utilize low‐frequency information synthesized by neural networks to obtain low‐wavenumber initial models. Previous attempts to reconstruct synthetic low‐frequency data using supervised learning methods have shown feasibility but were limited to training with synthetic data that required labeled information. In this study, we employed an unsupervised learning method, namely cycle‐consistent adversarial networks (CycleGAN), to reconstruct large‐scale‐feature related low‐frequency information based on the high‐frequency input data. Unlike supervised learning, CycleGAN allows the use of field data as input to train the network, which is more closely aligned with practical applications. Nevertheless, this approach presents challenges in terms of training complexity and potential output stability. To overcome these challenges, we reconstructed an appropriate target data set that combines high, medium, and low‐frequency components and incorporated additional loss functions to enhance the network's output performance. We conducted quantitative evaluations of the method's sensitivity to the target data set and its ability to handle low‐quality input data through numerical testing. The final results from field data testing confirmed the feasibility and effectiveness of the proposed method. Plain Language Summary: Full waveform inversion (FWI) is a state of art imaging technology, but missing low‐frequency energy in active seismic waveform leads to FWI failure on field data. To address this challenge, we used cycle‐consistent adversarial networks (CycleGAN), an unsupervised learning method, to generate missing low‐frequency information from high‐frequency field data. Unlike supervised learning methods that require labeled synthetic data, CycleGAN can directly use real‐world field data for training, making it more practical. We enhanced the method by constructing a suitable target data set to improve the network's performance. The field data test confirmed the effectiveness of our approach in reducing mismatch caused by large discrepancies between observed and synthesized data and improving subsurface imaging. Key Points: Low‐frequency data can help alleviate the issue of cycle‐skipping in full‐waveform inversionWe utilize unsupervised learning with field data as input to train the network for predicting relevant low‐frequency informationSwitching from a single low‐frequency target data set to a composite high‐medium‐low frequency form improves the network's performance [ABSTRACT FROM AUTHOR]

Artificial intelligence (AI) seismology has witnessed enormous success in a variety of fields, especially in earthquake detection and P and S‐wave arrival picking. It has become widely accepted that DL techniques greatly help routine seismic monitoring by enabling more accurate picking than traditional pickers like STA/LTA. However, a completely automatic AI‐driven earthquake monitoring framework has not been reported due to the concerns of potential false positives using DL pickers. Here, we propose a novel AI‐facilitated near real‐time monitoring framework using a recently developed deep learning (DL) picker (EQCCT) that has been deployed in the Texas seismological network (TexNet). For the West Texas area, TexNet's seismic monitoring relies on the EQCCT picker to report earthquake events. For earthquakes with a magnitude above two, the picks are further validated by analysts to output the final TexNet catalog. Due to the fast‐increasing seismicity caused by continuing oil&gas production in West Texas, this AI‐facilitated framework significantly relieves the workload of TexNet analysts. We show the mean absolute error (MAE) of automatic magnitude estimation for the magnitude‐above‐two earthquakes is smaller than 0.15 in West Texas and MAEs of hypocenter locations within 2.6 km in both distance and depth estimates. This research provides more evidence that DL pickers can play a fundamental role in daily earthquake monitoring. Plain Language Summary: AI has been a significant part of almost every aspect of our life. However, it is not widely accepted that AI can be a game changer in geoscience. In this work, we provide a compelling example that AI can significantly lower the overload of earthquake analysts in everyday work and boost our earthquake detectability, thereby enhancing our understanding of earthquake statistics, physics, and potential nucleation mechanisms. In a nutshell, AI has already been demonstrated to be successful in earthquake analysis but never in real‐time monitoring, which requires a high success rate and zero tolerance to large‐earthquake detection errors. Here, we show that AI‐assisted earthquake monitoring workflow can be almost 100% accurate for moderate‐to‐large earthquakes. Key Points: We propose an AI‐facilitated near real‐time monitoring framework that has been deployed at TexNet using a recently developed deep learning (DL) pickerDue to the fast‐increasing seismicity in West Texas, this AI‐facilitated framework significantly relieves the workload of TexNet analystsThe mean absolute error (MAE) of automatic magnitude estimation for the magnitude‐above‐two earthquakes is smaller than 0.15 in West Texas [ABSTRACT FROM AUTHOR]

Highly sensitive laser strainmeters at Moxa Geodynamic Observatory (MGO) measure motions of the upper Earth's crust. Since the mountain overburden of the laser strainmeters installed in the gallery of the observatory is relatively low, the recorded time series are strongly influenced by local meteorological phenomena. To estimate the nonlinear effect of the meteorological variables on strain measurements in a non‐stationary environment, advanced methods capable of learning the nonlinearity and discovering causal relationships in the non‐stationary multivariate tectonic‐climate time series are needed. Methods for causal inference generally perform well in identifying linear causal relationships but often struggle to retrieve complex nonlinear causal structures prevalent in real‐world systems. This work presents a novel model invariance‐based causal discovery (CDMI) method that utilizes deep networks to model nonlinearity in a multivariate time series system. We propose to use the theoretically well‐established Knockoffs framework to generate in‐distribution, uncorrelated copies of the original data as interventional variables and test the model invariance for causal discovery. To deal with the non‐stationary behavior of the tectonic‐climate time series recorded at the MGO, we propose a regime identification approach that we apply before causal analysis to generate segments of time series that possess locally consistent statistical properties. First, we evaluate our method on synthetically generated time series by comparing it to other causal analysis methods. We then investigate the hypothesized effect of meteorological variables on strain measurements. Our approach outperforms other causality methods and provides meaningful insights into tectonic‐climate causal interactions. Plain Language Summary: At the Moxa Geodynamic Observatory, highly sensitive laser devices are used to measure movements in the Earth's crust. However, these measurements can be influenced by local weather conditions because the equipment is not buried under a lot of soil and rock. To make sense of these measurements, we need methods that can understand how weather conditions affect the Earth's movements, especially when the relationship is complex. Here, we introduce a causal discovery method (CDMI) that uses advanced AI methods and statistical tests to uncover the complex causal relationships between climate and Earth's movements. Since both climate and Earth's conditions can change significantly over time, we also segment the data into different periods where changes are relatively small. This improves the accuracy of the overall causal analysis. We tested our method with artificial data and found that it performs better than other methods. Additionally, we applied our method to real data to understand how climate affects Earth's movements, gaining valuable insights into these complex interactions. Key Points: Long time series of laser‐strainmeter recordings at Geodynamic Observatory Moxa exhibit influences from environmental phenomenaTo understand the environmental influences on strainmeter recordings, a method is needed to identify causal relationshipsA deep learning causal inference method is presented to uncover causal relations in the nonlinear, non‐stationary multivariate time series [ABSTRACT FROM AUTHOR]

The ever-improving performances of physics-based simulations and the rapid developments of deep learning are offering new perspectives to study earthquake-induced ground motion. Due to the large amount of data required to train deep neural networks, applications have so far been limited to recorded data or two-dimensional (2D) simulations. To bridge the gap between deep learning and high-fidelity numerical simulations, this work introduces a new database of physics-based earthquake simulations. The HEterogeneous Materials and Elastic Waves with Source variability in 3D (HEMEW S -3D) database comprises 30 000 simulations of elastic wave propagation in 3D geological domains. Each domain is parametrized by a different geological model built from a random arrangement of layers augmented by random fields that represent heterogeneities. Elastic waves originate from a randomly located pointwise source parametrized by a random moment tensor. For each simulation, ground motion is synthesized at the surface by a grid of virtual sensors. The high frequency of waveforms (fmax⁡=5 Hz) allows for extensive analyses of surface ground motion. Existing and foreseen applications range from statistical analyses of the ground motion variability and machine learning methods on geological models to deep-learning-based predictions of ground motion that depend on 3D heterogeneous geologies and source properties. Data are available at 10.57745/LAI6YU. [ABSTRACT FROM AUTHOR]

Earthquake early warning (EEW) is of great significance in mitigating seismic disasters. Traditional EEW algorithms, which are knowledge‐driven approaches, rely on seismologists' analysis. The limited intensity measures were extracted by seismologists from P‐wave signals. And there is considerable uncertainty for predicting epicentral distance, magnitude, peak ground acceleration (PGA), and peak ground velocity (PGV). Currently, data‐driven deep learning methods with the strong learning abilities do not consider knowledge information from seismologists in EEW; thus, there is unexplored potential in enhancing the performance of deep learning models for EEW. Here, we construct the Data‐knowledge driven Hybrid deep Learning network (DHLnet) for EEW using the waveform input, knowledge embedding, convolutional neural network and graph convolutional network, aiming to integrate knowledge information from knowledge‐driven methods and the strong learning ability of data‐driven deep learning methods, that is, improving the performance of EEW. For the same test data set, compared with knowledge‐driven methods and data‐driven deep learning models, we demonstrate that DHLnet enhances the timeliness and robustness in predicting the epicentral distance, magnitude, PGA, and PGV during 10 s time window following the arrival of P‐wave. Furthermore, to validate the generalization and robustness of the DHLnet in EEW, we applied the trained DHLnet to an independent data set, within first few seconds after an earthquake occurs, DHLnet can provide robust magnitude estimation, epicentral distance estimation and high alarm accuracy. The potential of the proposed network is to enhance the performance of EEW systems and provides new insights into the exploration of deep learning methods for EEW domain. Plain Language Summary: Earthquake early warning (EEW) relies heavily on crucial parameters like epicentral distance, magnitude, and peak ground motion (peak ground acceleration [PGA] and velocity [PGV]). To quickly and accurately determine these parameters, traditional EEW algorithms, which are knowledge‐driven approaches, rely on seismologists' analysis based on earthquake rupture physics, and establish empirical EEW parameter prediction equations. Currently, data‐driven deep learning models with strong learning ability in EEW are mainly used to extract features from raw seismic waveforms, and do not take existing knowledge information from traditional EEW algorithms into account. Therefore, there is underutilized potential in improving the generalization, reliability and interpretability of deep learning model for EEW. Here, a Data‐knowledge driven Hybrid deep Learning network (DHLnet) for EEW is proposed and demonstrate that compared with knowledge‐driven methods and data‐driven deep learning models, DHLnet has better performance on predicting epicentral distance, magnitude, PGA and PGV. Key Points: A Data‐knowledge driven Hybrid deep Learning network (DHLnet) is proposed for earthquake early warning (EEW)The DHLnet mainly consists of the knowledge embedding, convolutional neural network and graph convolutional networkWe demonstrate that DHLnet outperforms knowledge‐driven EEW methods and data‐driven deep learning models [ABSTRACT FROM AUTHOR]

SUMMARY: Applications of machine learning in seismology have greatly improved our capability of detecting earthquakes in large seismic data archives. Most of these efforts have been focused on continental shallow earthquakes, but here we introduce an integrated deep-learning-based workflow to detect deep earthquakes recorded by a temporary array of ocean-bottom seismographs (OBSs) and land-based stations in the Tonga subduction zone. We develop a new phase picker, PhaseNet-TF, to detect and pick P- and S-wave arrivals in the time–frequency domain. The frequency-domain information is critical for analysing OBS data, particularly the horizontal components, because they are contaminated by signals of ocean-bottom currents and other noise sources in certain frequency bands. PhaseNet-TF shows a much better performance in picking S waves at OBSs and land stations compared to its predecessor PhaseNet. The predicted phases are associated using an improved Gaussian Mixture Model Associator GaMMA-1D and then relocated with a double-difference package teletomoDD. We further enhance the model performance with a semi-supervised learning approach by iteratively refining labelled data and retraining PhaseNet-TF. This approach effectively suppresses false picks and significantly improves the detection of small earthquakes. The new catalogue of Tonga deep earthquakes contains more than 10 times more events compared to the reference catalogue that was analysed manually. This deep-learning-enhanced catalogue reveals Tonga seismicity in unprecedented detail, and better defines the lateral extent of the double-seismic zone at intermediate depths and the location of four large deep-focus earthquakes relative to background seismicity. It also offers new potential for deciphering deep earthquake mechanisms, refining tomographic models, and understanding of subduction processes. [ABSTRACT FROM AUTHOR]

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

Recent geophysical studies have highlighted the potential utility of integrating both seismic and infrasound data to improve source characterization and event discrimination efforts. However, the influence of each of these data types within an integrated framework is not yet well‐understood by the geophysical community. To help elucidate the role of each data type within a merged structure, we develop a neural network which fuses seismic and infrasound array data via a gated multimodal unit for earthquake‐explosion discrimination within the Korean Peninsula. Model performance is compared before and after adding the infrasound branch. We find that the seismoacoustic model outperforms the seismic model, with the majority of the improvements stemming from the explosions class. The influence of infrasound is quantified by analyzing gated multimodal activations. Results indicate that the model relies comparatively more on the infrasound branch to correct seismic predictions. Plain Language Summary: Earthquakes and explosions can produce energy that travel as waves through the ground, seismic, and the air, infrasound. As these waves travel to the station where they are detected, they can be changed so drastically by the medium that it makes it difficult to determine what caused them. In these instances, it has been shown that using both seismic and infrasound data works better to characterize an event than using them independent of one another. However, due to the differences in how the air and ground influence the movement of energy, it is not well‐known how these types of data work in unison to give us more information about an event. In this study, we use a machine learning model trained on both seismic and infrasound data to help us better understand how they can be used together to determine their source. Key Points: Discrimination performance within the Korean Peninsula is improved after fusing seismoacoustic data within a deep learning architectureNeural network framework provides insight into how information in multimodal data combine to distinguish between different event types [ABSTRACT FROM AUTHOR]

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

The contagious and pandemic COVID-19 disease is currently considered as the main health concern and posed widespread panic across human-beings. It affects the human respiratory tract and lungs intensely. So that it has imposed significant threats for premature death. Although, its early diagnosis can play a vital role in revival phase, the radiography tests with the manual intervention are a time-consuming process. Time is also limited for such manual inspecting of numerous patients in the hospitals. Thus, the necessity of automatic diagnosis on the chest X-ray or the Computed Tomography (CT) images with a high effcient performance is urgent. Toward this end, we propose a novel method, named as the ULGFBP-ResNet51 to tackle with the COVID-19 diagnosis in the images. In fact, this method includes Uniform Local Binary Pattern (ULBP), Gabor Filter (GF), and 51-layer Residual Neural Networks (ResNet51). According to our results, this method could offer superior performance in comparison with the other methods, and attain maximum accuracy. [ABSTRACT FROM AUTHOR]

The application of deep‐learning‐based seismic phase pickers has surged in recent years. However, the efficacy of these models when applied to monitoring volcano seismicity has yet to be fully evaluated. Here, we first compile a data set of seismic waveforms from various volcanoes globally. We then show that the performances of two widely used deep‐learning pickers deteriorate systematically as the earthquakes' frequency content decreases. Therefore, the performances are especially poor for long‐period earthquakes often associated with fluid/magma movement. Subsequently, we train new models which perform significantly better, including when tested on two data sets where no training data were used: volcanic earthquakes along the Cascadia subduction zone and tectonic low‐frequency earthquakes along the Nankai Trough. Our model/workflow can be applied to improve monitoring of volcano seismicity globally while our compiled data set can be used to benchmark future methods for characterizing volcano seismicity, especially long‐period earthquakes which are difficult to monitor. Plain Language Summary: Earthquake activity at volcanic regions is often monitored to indicate volcanic activity. Identifying the time when the energy radiated from an earthquake source arrives at a seismometer is essential for locating the earthquake, which can be difficult for volcanic earthquakes because of high noise levels, high event rates, and obscured onsets. Previous studies have demonstrated that deep learning can excel in picking the arrival times of regular earthquakes. However, it is unclear how sensitive these detectors are to earthquakes in volcanic regions. Here, we first compile a data set of earthquakes from various volcanoes globally. We then show that existing deep‐learning‐based detectors can miss a large fraction of these earthquakes, especially those without an abrupt change in signal amplitude. We then provide two new models which can better detect volcanic earthquakes than existing models. Our model/workflow can be applied to improve monitoring of volcanic earthquakes globally. Key Points: We compile a data set of seismic waveforms from various volcanic regions globallyWe show that existing deep‐learning phase pickers' performances deteriorate with decreasing earthquake frequency contentOur retrained models perform better and are more generalizable for monitoring volcano seismicity, especially long‐period earthquakes [ABSTRACT FROM AUTHOR]

Estimating an earthquake's magnitude and location may not be necessary to predict shaking in real time; instead, wavefield‐based approaches predict shaking with few assumptions about the seismic source. Here, we introduce GRAph Prediction of Earthquake Shaking (GRAPES), a deep learning model trained to characterize and propagate earthquake shaking across a seismic network. We show that GRAPES' internal activations, which we call "seismic vectors", correspond to the arrival of distinct seismic phases. GRAPES builds upon recent deep learning models applied to earthquake early warning by allowing for continuous ground motion prediction with seismic networks of all sizes. While trained on earthquakes recorded in Japan, we show that GRAPES, without modification, outperforms the ShakeAlert earthquake early warning system on the 2019 M7.1 Ridgecrest, CA earthquake. Plain Language Summary: Have you ever heard something through the grapevine? It often takes you by surprise to hear a message from someone other than the original source. You might have felt an earthquake in a similar way: experiencing shaking (the message) at your location rather than movement along a fault (the source). We apply grapevine‐style communication to earthquake early warning (EEW). The goal of EEW is to warn people to prepare for earthquake shaking before damaging seismic waves arrive at their location. We build on recent work that used deep learning and large earthquake data sets to predict earthquake shaking. We developed a deep learning algorithm called GRAPES which predicts shaking in a manner similar to a game of seismic telephone: when a seismic sensor detects shaking, it sends a message to its neighboring sensors, warning them to expect shaking soon. These sensors then pass on the message to their more distant neighbors along the grapevine. We show that the messages GRAPES learned to send between sensors are like seismic status updates: "I'm seeing this type of seismic wave right now". We applied GRAPES to the 2019 M7.1 Ridgecrest, CA earthquake and it predicted shaking accurately and quickly. Key Points: A deep learning network trained to predict ground motion learned an internal representation of the seismic wavefieldIndividual neurons within the network activate with the arrival of P waves, S waves, surface waves, coda waves, and ambient noiseWhile trained on earthquakes in Japan, the model generalizes well to predicting ground motions for the 2019 Ridgecrest, CA earthquake [ABSTRACT FROM AUTHOR]

The Shear wave (S-wave) velocity is an essential parameter in reservoir characterization and evaluation, fluid identification, and prestack inversion. However, the cost of obtaining S-wave velocities directly from dipole acoustic logging is relatively high. At the same time, conventional data-driven S-wave velocity prediction methods exhibit several limitations, such as poor accuracy and generalization of empirical formulas, inadequate exploration of logging curve patterns of traditional fully connected neural networks, and gradient explosion and gradient vanishing problems of recurrent neural networks (RNNs). In this study, we present a reliable and low-cost deep learning (DL) approach for S-wave velocity prediction from real logging data to facilitate the solution of these problems. We designed a new network sensitive to depth sequence logging data using conventional neural networks. The new network is composed of one-dimensional (1D) convolutional, bidirectional long short-term memory (BiLSTM), attention, and fully connected layers. First, the network extracts the local features of the logging curves using a 1D convolutional layer, and then extracts the long-term sequence features of the logging curves using the BiLSTM layer, while adding an attention layer behind the BiLSTM network to further highlight the features that are more significant for S-wave velocity prediction and minimize the influence of other features to improve the accuracy of S-wave velocity prediction. Afterward, the nonlinear mapping relationship between logging data and S-wave velocity is established using several fully connected layers. We applied the new network to real field data and compared its performance with three traditional methods, including a long short-term memory (LSTM) network, a back-propagation neural network (BPNN), and an empirical formula. The performance of the four methods was quantified in terms of their coefficient of determination (R²), root mean square error (RMSE), and mean absolute error (MAE). The new network exhibited better performance and generalization ability, with R² greater than 0.95 (0.9546, 0.9752, and 0.9680, respectively), RMSE less than 57 m/s (56.29, 23.18, and 30.17 m/s, respectively), and MAE less than 35 m/s (34.68, 16.49, and 21.47 m/s, respectively) for the three wells. The test results demonstrate the efficacy of the proposed approach, which has the potential to be widely applied in real areas where S-wave velocity logging data are not available. Furthermore, the findings of this study can help for a better understanding of the superiority of deep learning schemes and attention mechanisms for logging parameter prediction. [ABSTRACT FROM AUTHOR]

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]

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

Skin cancer is a significant health concern worldwide, and early and accurate diagnosis plays a crucial role in improving patient outcomes. In recent years, deep learning models have shown remarkable success in various computer vision tasks, including image classification. In this research study, we introduce an approach for skin cancer classification using vision transformer, a state-of-the-art deep learning architecture that has demonstrated exceptional performance in diverse image analysis tasks. The study utilizes the HAM10000 dataset; a publicly available dataset comprising 10,015 skin lesion images classified into two categories: benign (6705 images) and malignant (3310 images). This dataset consists of high-resolution images captured using dermatoscopes and carefully annotated by expert dermatologists. Preprocessing techniques, such as normalization and augmentation, are applied to enhance the robustness and generalization of the model. The vision transformer architecture is adapted to the skin cancer classification task. The model leverages the self-attention mechanism to capture intricate spatial dependencies and long-range dependencies within the images, enabling it to effectively learn relevant features for accurate classification. Segment Anything Model (SAM) is employed to segment the cancerous areas from the images; achieving an IOU of 96.01% and Dice coefficient of 98.14% and then various pretrained models are used for classification using vision transformer architecture. Extensive experiments and evaluations are conducted to assess the performance of our approach. The results demonstrate the superiority of the vision transformer model over traditional deep learning architectures in skin cancer classification in general with some exceptions. Upon experimenting on six different models, ViT-Google, ViT-MAE, ViT-ResNet50, ViT-VAN, ViT-BEiT, and ViT-DiT, we found out that the ML approach achieves 96.15% accuracy using Google's ViT patch-32 model with a low false negative ratio on the test dataset, showcasing its potential as an effective tool for aiding dermatologists in the diagnosis of skin cancer. [ABSTRACT FROM AUTHOR]

Seismic phase picking and magnitude estimation are fundamental aspects of earthquake monitoring and seismic event analysis. Accurate phase picking allows for precise characterization of seismic wave arrivals, contributing to a better understanding of earthquake events. Likewise, accurate magnitude estimation provides crucial information about an earthquake's size and potential impact. Together, these components enhance our ability to monitor seismic activity effectively. In this study, we explore the application of deep-learning techniques for earthquake detection and magnitude estimation using continuous seismic recordings. Our approach introduces DynaPicker, which leverages dynamic convolutional neural networks to detect seismic body-wave phases in continuous seismic data. We demonstrate the effectiveness of DynaPicker using various open-source seismic datasets, including both window-format and continuous recordings. We evaluate its performance in seismic phase identification and arrival-time picking, as well as its robustness in classifying seismic phases using low-magnitude seismic data in the presence of noise. Furthermore, we integrate the phase arrival-time information into a previously published deep-learning model for magnitude estimation. We apply this workflow to continuous recordings of aftershock sequences following the Turkey earthquake. The results of this case study showcase the reliability of our approach in earthquake detection, phase picking, and magnitude estimation, contributing valuable insights to seismic event analysis. [ABSTRACT FROM AUTHOR]

Myanmar bears a high risk of destructive earthquakes, yet detailed seismicity catalogs are rare. We designed a deep‐learning‐based data processing pipeline and applied it to the data recorded by a large‐aperture (∼400 km) seismic array in central Myanmar to produce a high‐resolution earthquake catalog. We precisely located 1891 earthquakes at shallow (<50 km) depth, a 2‐fold increase compared to the traditional procedures. The new catalog reveals the Kabaw Fault seismicity disappears south of ∼22.8°N, where the deeper (20–40 km) seismicity appears west of the southern Kabaw Fault. Such seismicity contrast along the strike of the Kabaw Fault possibly implies an along‐strike change of deformation responses to the shortening process by the India plate oblique subduction. The middle segment of the Sagaing Fault is likely locked and prone to hosting large earthquakes according to the derived low b‐value. Plain Language Summary: Myanmar is a highly seismically active region, yet fault geometry and activities remain poorly understood because of limited modern seismological investigations. Here, we designed a set of machine‐learning algorithms to detect small earthquakes and determine their locations precisely. The seismic data are recorded by a temporary seismic network deployed in central Myanmar. We obtained twice as many earthquakes as the previous research used the regular procedure. Our improved earthquake data set unveils seismic activity changes along the Kabaw Fault through the changes in earthquake locations, depths, and magnitude‐frequency relations. Kabaw Fault is an import boundary fault in the subduction system of the Indo‐Burma Range. This subtle change was not previously observed but means a significant alternation in deformation style along the subduction strike. Moreover, our improved data set indicates that the Sagaing Fault, the most active fault in Myanmar, is prone to generating large earthquakes in the future. This implication warns the nearby populated cities, like Mandalay, of a significant megaquake threat. Key Points: We detect 1891 shallow earthquakes in Myanmar with a deep‐learning‐empowered pipeline, a 2‐fold increase against the routine procedureN‐S seismicity discrepancy is observed near the Kabaw Fault and may imply different responses to E‐W shortening by the Indian subductionLow b‐value derived from the new catalog on the middle Sagaing Fault indicates a high risk of destructive earthquakes [ABSTRACT FROM AUTHOR]

Data reconstruction is the most essential step in seismic data processing. Although the compressed sensing (CS) theory breaks through the Nyquist sampling theorem, we previously proved that the CS-based reconstruction of spatially irregular seismic data could not fully meet the theoretical requirements, resulting in low reconstruction accuracy. Although deep learning (DL) has great potential in mining features from data and accelerating the process, it faces challenges in earth science such as limited labels and poor generalizability. To improve the generalizability of deep neural network (DNN) in reconstructing seismic data in the actual situation of limited labeling, this paper proposes a method called CSDNN that combines model-driven CS and data-driven DNN to reconstruct the spatially irregular seismic data. By physically constraining neural networks, this method increases the generalizability of the network and improves the insufficient reconstruction caused by the inability to sample randomly in the whole data definition domain. Experiments on the synthetic and field seismic data show that the CSDNN reconstruction method achieves better performance compared with the conventional CS method and DNN method, including those with low sampling rates, which verifies the feasibility, effectiveness and generalizability of this approach. [ABSTRACT FROM AUTHOR]

Satellite‐based optical video sensors are poised as the next frontier in remote sensing. Satellite video offers the unique advantage of capturing the transient dynamics of floods with the potential to supply hitherto unavailable data for the assessment of hydraulic models. A prerequisite for the successful application of hydraulic models is their proper calibration and validation. In this investigation, we validate 2D flood model predictions using satellite video‐derived flood extents and velocities. Hydraulic simulations of a flood event with a 5‐year return period (discharge of 722 m3 s−1) were conducted using Hydrologic Engineering Center—River Analysis System 2D in the Darling River at Tilpa, Australia. To extract flood extents from satellite video of the studied flood event, we use a hybrid transformer‐encoder, convolutional neural network (CNN)‐decoder deep neural network. We evaluate the influence of test‐time augmentation (TTA)—the application of transformations on test satellite video image ensembles, during deep neural network inference. We employ Large Scale Particle Image Velocimetry (LSPIV) for non‐contact‐based river surface velocity estimation from sequential satellite video frames. When validating hydraulic model simulations using deep neural network segmented flood extents, critical success index peaked at 94% with an average relative improvement of 9.5% when TTA was implemented. We show that TTA offers significant value in deep neural network‐based image segmentation, compensating for aleatoric uncertainties. The correlations between model predictions and LSPIV velocities were reasonable and averaged 0.78. Overall, our investigation demonstrates the potential of optical space‐based video sensors for validating flood models and studying flood dynamics. Plain Language Summary: Videos of the Earth surface recorded by satellites can enable us to observe and characterize dynamic moving features, such as floods, that would otherwise be very difficult or dangerous to investigate from the ground. Hydrologists often rely on using physics‐based computer models to simulate flood events, but require observational data to make sure these reflect reality accurately. We use artificial intelligence techniques to automatically detect flood extents from satellite video, and track surface features from frame to frame in order to measure how fast the water surface is flowing. Satellite video was collected during opportunistically clear skies in January 2022, along a 6.5 km length of the River Darling in Australia. The flood extent and flow velocities were used to improve numerical model predictions of the flood event. Our findings demonstrate the considerable promise of satellite video to complement existing flood mapping and modeling approaches, and to provide insight into the earth's hydrosphere, particularly in remote locations and during extreme conditions. Key Points: Satellite video derived flood extents and velocities successfully validate 2D hydraulic model predictionsTest‐time augmentation during deep learning inference improved flood extent delineation and enhanced 2D model validation metricsIncorporating characterization of discharge uncertainty into hydraulic model predictions resulted in more accurate model validation [ABSTRACT FROM AUTHOR]

Detecting phase arrivals and pinpointing the arrival times of seismic phases in seismograms is crucial for many seismological analysis workflows. For land station data, machine learning methods have already found widespread adoption. However, deep learning approaches are not yet commonly applied to ocean bottom data due to a lack of appropriate training data and models. Here, we compiled an extensive and labeled ocean bottom seismometer (OBS) data set from 15 deployments in different tectonic settings, comprising ∼90,000 P and ∼63,000 S manual picks from 13,190 events and 355 stations. We propose PickBlue, an adaptation of the two popular deep learning networks EQTransformer and PhaseNet. PickBlue joint processes three seismometer recordings in conjunction with a hydrophone component and is trained with the waveforms in the new database. The performance is enhanced by employing transfer learning, where initial weights are derived from models trained with land earthquake data. PickBlue significantly outperforms neural networks trained with land stations and models trained without hydrophone data. The model achieves a mean absolute deviation of 0.05 s for P‐waves and 0.12 s for S‐waves, and we apply the picker on the Hikurangi Ocean Bottom Tremor and Slow Slip OBS deployment offshore New Zealand. We integrate our data set and trained models into SeisBench to enable an easy and direct application in future deployments. Plain Language Summary: Ocean bottom seismometers (OBS) are seismic stations on the seafloor. Just like their counterparts on land, they record many earthquakes on three component sensors but are additionally equipped with underwater hydrophones. To determine the location of an earthquake, seismologists must precisely measure the arrival times of seismic waves. For onshore data, machine learning (ML) has been highly successful in determining earthquake arrival times. However, the noise and the signal are different in the ocean environment. For example, the recordings can contain whale songs and water layer reverberations and are disturbed by ocean bottom currents. We have assembled an extensive database of ocean bottom recordings and trained artificial neural networks to use the underwater hydrophone information and cope with the ocean noise environment. We demonstrate that the resulting ML picker picks are similar to those of human experts and outperform phase pickers based on land data only. We compare earthquake catalogs based on different pickers created from an OBS deployment offshore New Zealand and demonstrate that PICKBLUE outperforms previous pickers. We make the database and ML picker available with a standard interface so that it is easy for other scientists to apply them in their studies. Key Points: We assembled a database of ocean Bottom Seismometer (OBS) waveforms and manual P and S picks, on which we train PickBlue, a deep learning pickerOur picker significantly outperforms pickers trained with land‐based data with confidence values reflecting the likelihood of outlier picksThe picker and database are available in the SeisBench platform, allowing easy and direct application to OBS traces and hydrophone records [ABSTRACT FROM AUTHOR]

Seismic wave arrival time measurements form the basis for numerous downstream applications. State‐of‐the‐art approaches for phase picking use deep neural networks to annotate seismograms at each station independently, yet human experts annotate seismic data by examining the whole network jointly. Here, we introduce a general‐purpose network‐wide phase picking algorithm based on a recently developed machine learning paradigm called Neural Operator. Our model, called Phase Neural Operator, leverages the spatio‐temporal contextual information to pick phases simultaneously for any seismic network geometry. This results in superior performance over leading baseline algorithms by detecting many more earthquakes, picking more phase arrivals, while also greatly improving measurement accuracy. Following similar trends being seen across the domains of artificial intelligence, our approach provides but a glimpse of the potential gains from fully‐utilizing the massive seismic data sets being collected worldwide. Plain Language Summary: Earthquake monitoring often involves measuring arrival times of P‐ and S‐waves of earthquakes from continuous seismic data. With the advancement of artificial intelligence, state‐of‐the‐art phase picking methods use deep neural networks to examine seismic data from each station independently; this is in stark contrast to the way that human experts annotate seismic data, in which waveforms from the whole network containing multiple stations are examined simultaneously. With the performance gains of single‐station algorithms approaching saturation, it is clear that meaningful future advances will require algorithms that can naturally examine data for entire networks at once. Here we introduce a multi‐station phase picking algorithm based on a recently developed machine learning paradigm called Neural Operator. Our algorithm, called Phase Neural Operator, leverages the spatial‐temporal information of earthquake signals from an input seismic network with arbitrary geometry. This results in superior performance over leading baseline algorithms by detecting many more earthquakes, picking many more seismic wave arrivals, yet also greatly improving measurement accuracy. Key Points: We introduce a multi‐station phase picking algorithm, Phase Neural Operator (PhaseNO), that is based on a new machine learning paradigm called Neural OperatorPhaseNO can use data from any number of stations arranged in any arbitrary geometry to pick phases across the input stations simultaneouslyBy leveraging the spatial and temporal contextual information, PhaseNO achieves superior performance over leading baseline algorithms [ABSTRACT FROM AUTHOR]

Deep-learning (DL) pickers have demonstrated superior performance in seismic phase picking compared to traditional pickers. DL pickers are extremely effective in processing large amounts of seismic data. Nevertheless, they encounter challenges when handling seismograms from different tectonic environments or source types, and even a slight change in the input waveform can considerably affect their consistency. Here, we fine-tuned a self-trained deep neural network picker using a small amount of local seismic data (26,875 three-component seismograms) recorded by regional seismic networks in South Korea. The self-trained model was developed using publicly available waveform datasets, comprising over two million three-component seismograms. The results revealed that the Korean-fine-tuned phase picker (KFpicker) effectively enhanced picking quality, even when applied to data that were not used during the fine-tuning process. When compared to the performance of the pre-trained model, this improvement was consistently observed regardless of variations in the positions of seismic phases in the input waveform, Furthermore, when the KFpicker predicted the phases for overlapping input windows and used the median value of probabilities as a threshold for phase detection, a considerable decrease was observed in the number of false picks. These findings indicate that fine-tuning a deep neural network using a small amount of local data can improve earthquake detection in the region of interest, while careful data augmentation can enhance the robustness of DL pickers against variations in the input window. The application of KFpicker to the 2016 Gyeongju earthquake sequence yielded approximately twice as many earthquakes compared to previous studies. Consequently, detailed and instantaneous statistical parameters of seismicity can be evaluated, making it possible to assess seismic hazard during an earthquake sequence. [ABSTRACT FROM AUTHOR]

Distributed Acoustic Sensing (DAS) is an emerging technology for earthquake monitoring and subsurface imaging. However, its distinct characteristics, such as unknown ground coupling and high noise level, pose challenges to signal processing. Existing machine learning models optimized for conventional seismic data struggle with DAS data due to its ultra-dense spatial sampling and limited manual labels. We introduce a semi-supervised learning approach to address the phase-picking task of DAS data. We use the pre-trained PhaseNet model to generate noisy labels of P/S arrivals in DAS data and apply the Gaussian mixture model phase association (GaMMA) method to refine these noisy labels and build training datasets. We develop PhaseNet-DAS, a deep learning model designed to process 2D spatio-temporal DAS data to achieve accurate phase picking and efficient earthquake detection. Our study demonstrates a method to develop deep learning models for DAS data, unlocking the potential of integrating DAS in enhancing earthquake monitoring. In this study, the authors develop a semi-supervised approach to train a deep learning model, PhaseNet-DAS, for identifying seismic phases in Distributed Acoustic Sensing (DAS) data, which enables detecting and locating earthquakes using fiber-optic networks. [ABSTRACT FROM AUTHOR]

The Hindu Kush-Pamir region (HKPR) is characterized by complex ongoing deformation, unique slab geometry, and intermediate seismic activity. The availability of extensive seismological data in recent decades has prompted the use of deep learning algorithms to extract valuable insights. In this study, we present a fully automated approach for augmenting earthquake catalogue within the HKPR. Our method leverages an attention mechanism-based deep learning architecture to simultaneously detect events, perform phase picking, and estimate magnitudes. We applied this model to a ten-month dataset (January 2013–October 2013) from 83 stations in the region. Utilizing a robust criterion to evaluate the model’s probabilities, we associated phases at different stations and pinpointed earthquake locations in the region. Our results demonstrate a significant enhancement, revealing nearly four and a half times more earthquakes than previously documented in the International Seismological Center (ISC) catalogue. A notable portion of these newly detected events falls within the category of very low-magnitude earthquakes (<3), which were absent in the ISC catalogue. Notably, our spatiotemporal analysis reveals a concentration of crustal seismicity along poorly mapped neotectonic north and northeast oriented faults in the western Pamir, as well as the Vakhsh Thrust System and the Darvaz Karakul Fault. These findings underscore potential sources of future seismic hazards. Furthermore, our expanded earthquake catalogue facilitates a deeper understanding of the interplay between crustal and intermediate seismic activity in the HKPR, shedding light on the deformation and active faulting resulting from Eurasian-Indian plate interactions. [ABSTRACT FROM AUTHOR]

Earthquake early warning (EEW) systems aim to forecast the shaking intensity rapidly after an earthquake occurs and send warnings to affected areas before the onset of strong shaking. The system relies on rapid and accurate estimation of earthquake source parameters. However, it is known that source estimation for large ruptures in real‐time is challenging, and it often leads to magnitude underestimation. In a previous study, we showed that machine learning, HR‐GNSS, and realistic rupture synthetics can be used to reliably predict earthquake magnitude. This model, called Machine‐Learning Assessed Rapid Geodetic Earthquake model (M‐LARGE), can rapidly forecast large earthquake magnitudes with an accuracy of 99%. Here, we expand M‐LARGE to predict centroid location and fault size, enabling the construction of the fault rupture extent for forecasting shaking intensity using existing ground motion models. We test our model in the Chilean Subduction Zone with thousands of simulated and five real large earthquakes. The result achieves an average warning time of 40.5 s for shaking intensity MMI4+, surpassing the 34 s obtained by a similar GNSS EEW model. Our approach addresses a critical gap in existing EEW systems for large earthquakes by demonstrating real‐time fault tracking feasibility without saturation issues. This capability leads to timely and accurate ground motion forecasts and can support other methods, enhancing the overall effectiveness of EEW systems. Additionally, the ability to predict source parameters for real Chilean earthquakes implies that synthetic data, governed by our understanding of earthquake scaling, is consistent with the actual rupture processes. Plain Language Summary: Earthquake Early Warning (EEW) systems can evaluate the shaking impact soon after an earthquake occurs and send alerts to areas where strong shaking has not yet arrived. However, predicting shaking intensity for large earthquakes face challenges such as magnitude underestimation and rupture complexity, leading to inaccurate prediction. In our previous study, we proposed a machine learning model, called M‐LARGE, which can predict the magnitude of large earthquakes accurately and rapidly by utilizing surface deformation patterns recorded by HR‐GNSS. Here, we expand the M‐LARGE model to also predict the source location and fault size. This capability allows us to construct a fault, enabling forecasts of strong shaking 35–40 s before the shaking onset, longer than other similar methods. Our approach has the potential to improve the accuracy of EEW and reduce the impact of seismic hazards. Because the model is fully trained with simulated data, the successful prediction of real Chilean earthquakes indicates that our assumptions in simulating large earthquakes are sufficiently consistent with the rupture dynamics of actual events. Key Points: We train a ML model that utilizes HR‐GNSS data to predict source parameters and shaking intensity for large (M7.0+) earthquakesThe model yields a long median warning time of 35–40 s for MMI 4+ when tested on synthetic and real HR‐GNSS recordsOur model fills a critical gap in current EEW systems for large events and can operate in parallel with other methods for better warnings [ABSTRACT FROM AUTHOR]

We develop a novel approach for multi‐frequency, elliptical‐anisotropic eikonal tomography based on physics‐informed neural networks (pinnEAET). This approach simultaneously estimates the medium properties controlling anisotropic Rayleigh waves and reconstructs the traveltimes. The physics constraints built into pinnEAET's neural network enable high‐resolution results with limited inputs by inferring physically plausible models between data points. Even with a single source, pinnEAET can achieve stable convergence on key features where traditional methods lack resolution. We apply pinnEAET to ambient noise data from a dense seismic array (ChinArray‐Himalaya II) in the northeastern Tibetan Plateau with only 20 quasi‐randomly distributed stations as sources. Anisotropic phase velocity maps for Rayleigh waves in the period range from 10–40 s are obtained by training on observed traveltimes. Despite using only about 3% of the total stations as sources, our results show low uncertainties, good resolution and are consistent with results from conventional tomography. Plain Language Summary: Anisotropy refers to the directional dependence of seismic wave velocities, which can arise from a variety of factors such as crystal alignment, stress fields, or fluid‐filled cracks. Elliptical‐anisotropic eikonal tomography is a variant of eikonal tomography that can be used to estimate medium properties and reconstructed traveltimes from ambient noise data. In this study, we propose a new algorithm to implement multi‐frequency, elliptical‐anisotropic eikonal tomography based on physics‐informed neural networks (pinnEAET), which combine data‐driven models with theory‐based models that include physics constraints on the system. We apply this architecture to data from a dense seismic array deployed on the northeastern Tibetan Plateau. Our results can achieve at least the same resolution as traditional methods while requiring less traveltime data. This strategy can provide new insights into the seismic imaging in case of limited or noisy data. Key Points: We present a physics‐informed deep learning eikonal tomography method for anisotropic velocity modelingThe algorithm incorporates wave physics to simultaneously process multi‐frequency data, ensuring reliable tomographic modelsWe successfully recover the anisotropic velocity structure of the northeastern Tibet using less data than in traditional models [ABSTRACT FROM AUTHOR]

Seismic activity in the Northern Canadian Cordillera is characterized by diffuse earthquakes that extend hundreds of km northwest from the Yakutat collision zone. We use 25 months of broadband seismic data from Mackenzie Mountain Earthscope Project (MMEP), USArray Transportable Array (TA), and permanent Canadian National Seismic Network stations to present a local earthquake catalog with high sensitivity to small regional events. Deep learning techniques are adopted for both seismic phase detection and association. Event relocations are performed to provide well constrained estimates of earthquake depth distributions. Clusters of seismicity spanning the upper crust are located in the central Richardson Mountains, along the Tintina fault, and in the northeast Selwyn Basin. These clusters suggest that the core of the Richardson Anticlinorium is tectonically active and that the Tintina fault is a locus for low levels of active deformation. We interpret seismicity in the northeast Selwyn Basin as primarily occurring in the hanging wall of the Plateau thrust fault and suggest that some combination of localized duplex structures and lithological strength contrasts both within the Selwyn Basin and between abutting Paleozoic shelf sequences may be responsible for seismicity in the Mackenzie Mountain foreland. Plain Language Summary: The Northern Canadian Cordillera is situated inboard of a transition from a right‐lateral tectonic regime in coastal southeast Alaska to a subduction setting beneath southern Alaska. It presents a unique case study to examine the length scales over which tectonic deformation is able to permeate into a continental interior. In this study we use data from a relatively dense regional network of seismometers in order to present the most spatially complete catalog of earthquakes in the Northern Canadian Cordillera to date. We use machine learning techniques for the detection and classification of small magnitude earthquakes and use our results to analyze regional tectonics. Active faults that host earthquakes are identified and we use the earthquake depth distributions to further distinguish between mechanisms for regional stress transfer. We find evidence that the Tintina fault is presently active along a confined area at a low rate. We also find that low magnitude earthquakes do not generally occur within the Mackenzie Mountain Fold‐Thrust belt but are prevalent within the foreland Selwyn Basin. We also detect a pattern of relatively deep (∼30 km) earthquakes in the northern Richardson Mountains. Key Points: Small magnitude earthquakes extending to 25 km depth are detected throughout the Selwyn Basin in the Mackenzie Mountain forelandThe Tintina fault hosts a range of low magnitude earthquakes in the uppermost 15 km of crust within a 400 km zone along strikeDeformation in the Richardson Mountains is localized onto a narrow band of faults centered on the Richardson Anticlinorium [ABSTRACT FROM AUTHOR]

This article provides a review of gender classification using Kinect sensors in computer vision and biometrics. The article discusses the features used for gender classification, including facial, body, biological, and social network features. It also highlights the challenges and limitations of these features, such as diversity in gender identity and expression and biases in the training data. The article explores various methods and approaches for gender classification, including machine learning algorithms and statistical models. It emphasizes the importance of approaching gender classification with sensitivity and caution to avoid reinforcing harmful stereotypes or biases. The article concludes by mentioning the availability of public Kinect databases for research purposes. [Extracted from the article]

Geological hazard is an adverse geological condition that can cause loss of life and property. Accurate prediction and analysis of geological hazards is an important and challenging task. In the past decade, there has been a great expansion of geohazard detection data and advancement in data-driven simulation techniques. In particular, great efforts have been made in applying deep learning to predict geohazards. To understand the recent progress in this field, this paper provides an overview of the commonly used data sources and deep neural networks in the prediction of a variety of geological hazards. [ABSTRACT FROM AUTHOR]

An approximate 236‐year interval in major seismic activity near the southern Xianshuihe fault terminated on 5 September 2022 until a destructive M6.8 Luding earthquake. The strike‐slip mainshock, accompanied by potent normal‐faulting aftershocks, fell short of the previously anticipated M7+ event. Utilizing deep‐learning and source inversion techniques, we analyzed dense near‐field seismograms to examine detailed fault structures and rupture characteristics. Our high‐quality relocated catalog with 7,388 events revealed that the earthquake ruptured an unmapped multiscale fault network. The analysis of the mainshock and 43 ML ≥ 2.8 aftershocks suggests normal‐faulting events relate to the vertical movement of the Gongga Mountain. The sequence length, seismic gaps, and asperities derived from seismicity and waveform modeling imply that the rupture of the M6.8 quake was incomplete, indicating a high risk of a future M6+ event at the northern Moxi segment. These findings hold crucial implications for assessing future seismic hazards in this region. Plain Language Summary: On 5 September 2022, a destructive M6.8 earthquake struck the southern Xianshuihe fault, Luding, China. This event ended the 236‐year‐long silence in the area, resulting in 93 deaths and severe economic damage. This strike‐slip mainshock, accompanied by unusual normal‐faulting aftershocks, fell into earlier predictions of an approximate 248‐year cycle of M7+ earthquakes in this area. However, its magnitude was less than anticipated, leading us to question whether another significant earthquake could potentially occur. To investigate this, we used a novel deep‐learning method named DiTingPicker, which was trained with ∼5 million seismograms, and waveform modeling techniques to examine the detailed fault structures and mainshock rupture characteristics using dense local seismic stations. We compiled a 415‐day‐long earthquake catalog, showing this sequence ruptured an unmapped multiscale fault network of interlaced northwest‐trending and northeast‐trending faults. Our study suggests the normal‐faulting aftershocks are related to the vertical movement of the Gongga Mountain. By comparing the rupture lengths of the historical 1786 M7.8 earthquake and the 2022 event, our research suggests a potential risk for another M6+ earthquake at the unruptured northern Moxi segment. Key Points: A 415‐day seismic catalog that characterizes fine‐scale fault structures and seismic gaps was compiled using deep learningA hidden developing fault network was revealed and distinct normal‐fault aftershocks were spotted west of the strike‐slip mainshockSeismicity and source inversion suggest the M6.8 quake's rupture was incomplete and a future M6+ event at the unruptured Moxi segment [ABSTRACT FROM AUTHOR]

Seismology is witnessing explosive growth in the diversity and scale of earthquake catalogs. A key motivation for this community effort is that more data should translate into better earthquake forecasts. Such improvements are yet to be seen. Here, we introduce the Recurrent Earthquake foreCAST (RECAST), a deep‐learning model based on recent developments in neural temporal point processes. The model enables access to a greater volume and diversity of earthquake observations, overcoming the theoretical and computational limitations of traditional approaches. We benchmark against a temporal Epidemic Type Aftershock Sequence model. Tests on synthetic data suggest that with a modest‐sized data set, RECAST accurately models earthquake‐like point processes directly from cataloged data. Tests on earthquake catalogs in Southern California indicate improved fit and forecast accuracy compared to our benchmark when the training set is sufficiently long (>104 events). The basic components in RECAST add flexibility and scalability for earthquake forecasting without sacrificing performance. Plain Language Summary: We explore the potential for deep learning in earthquake forecasting. Prior work has relied heavily on statistical models that do not scale to fully utilize the currently available large earthquake data sets. Here we build on recent developments in deep learning for forecasting event sequences in general to create an implementation for earthquake data. The new approach allows us to incorporate larger data sets, potentially with more information about each earthquake. We also avoid a specific functional form, so the method naturally adapts to additional information about events, like magnitude or variations in behavior over time. As we add more data, results show continued improvements. This ability to incorporate and improve continually as training data sets increase indicates that there is more information in the earthquake catalogs than has yet been used for earthquake forecasting. Key Points: We introduce a deep learning model for earthquake forecasting and explore its performance on synthetic and regional earthquake data setsIt is flexible in the sense that a predefined functional form is not requiredIt is scalable in two senses: it is efficient on large data sets, and its performance relative to benchmarks improves with more training data [ABSTRACT FROM AUTHOR]

Breast cancer is cancer that develops from the breast tissue and has been recognized as one of the most dangerous and deadly diseases that is the second leading cause of cancer deaths in women. To help doctors and radiologists to diagnose these tumors as well as decrease the time and increase the accuracy, many machine learning methods have been implemented by now. Most of these methods suffer from extracting some significant features that represent the boundary of tumors. This is due to the fact that benign and malignant tumors can be considered the same if some borders cannot segment properly. So, in this study, we propose an automatic breast tumor segmentation and recognition based on a shallow convolutional neural network that uses multi-feature extraction routes. Also, an image enhancement approach is used before applying the image into the model which leads to avoiding a very deep structure. Our strategy leads to improvement in detecting the border of tumors and boosts the classification accuracy of tumors. We evaluated our pipeline on Mammographic Image Analysis Society (Mini-MIAS) and Digital Database for Screening Mammography (DDSM) datasets. The developed model can localize and classify tumors with the accuracy of 0.936, 0.890, 0.871 on the DDSM, and 0.944, 0.915, 0.892 on the Mini-MIAS, for normal, benign, and malignant regions, respectively. [ABSTRACT FROM AUTHOR]

Mine microseismic signal denoising is a basic and crucial link in microseismic data processing, which influences the accuracy and reliability of the monitoring system, and is of great significance with regard to safety during mining. Therefore, this study introduces a deep learning method to improve the mapping function and sparsity of signals in the time-frequency domain and constructs a denoising framework based on a deep convolutional autoencoder to address the denoising problem of mine microseismic signals. First, all noisy microseismic signals are normalized to ensure the nonlinear expression ability of the constructed denoising framework. Then, the normalized signals are transformed into the time-frequency domain using the short-time Fourier transform (STFT), and the real and imaginary parts of time-frequency coefficients serve as the input of the deep convolutional autoencoder to output the masks of the effective and noise signals. Next, these masks are applied to the time-frequency coefficients of the noisy microseismic signals, and the time-frequency coefficients of the potentially effective and noise signals are estimated. Finally, inverse STFT is used to transform these time-frequency coefficients to the time domain to obtain the final denoised effective and noise signals. The constructed framework automatically learns rich features from synthetic data to separate the effective and noise signals, thereby achieving the purpose of fast and automatic denoising. The experimental results show that compared with the wavelet threshold and ensemble empirical mode decomposition, the denoising framework considerably improves the signal-to-noise ratio of mine microseismic signals with less waveform distortion. Moreover, it can achieve a better denoising effect efficiently even in the case of a low SNR, which has obvious advantages. The constructed denoising framework is suitable for microseismic monitoring signals of various mine dynamic disasters and provides strong technical support for intelligent monitoring and early warning concerning production risks in mines. [ABSTRACT FROM AUTHOR]

Unknown geology ahead of the tunnel boring machine (TBM) brings a large safety risk for tunnel construction. Seismic aheadprospecting using TBM drilling noise as a source can achieve near-real-time detection, meeting the requirements of TBM rapid drilling. Seismic wavefield retrieval is the key data processing step for the efficient utilization of TBM drilling noise. The traditional solution is based on cross-correlation to extract reflected waves, but the reference waves remain in the result, disturbing the imaging and interpretation of the adverse geology. To solve this problem, the deep learning method was introduced in wavefield retrieval to improve the accuracy of geological prospecting. We trained a deep neural network (DNN) with its strong nonlinear mapping capability to transform seismic data from TBM drilling noise to data from the active source. The issue lies in its features for this specific tunnel task, including the decay of the seismic signal with time and the incomplete spatial correspondence. Thus, we improved a classical DNN with the time constraint as an additional input, and an additional pre-decoder to enlarge the receptive field. Additionally, a loss function weighted by the ground truth and time constraint is improved to achieve an accurate retrieval of the effective signal, considering the little effective information in tunnel data. Finally, the workflow of the proposed method was given, and a dataset designed with reference to the field case was employed to train the network. The proposed method accurately retrieved the reflection signal with higher dominant frequencies, which helped improve the accuracy of imaging. Numerical simulations and imaging on typical geological models show that the proposed method can suppress reference waves and get more accurate results with fewer artifacts. The proposed method has been applied in the Gaoligongshan Tunnel and imaged two abnormal zones, providing meaningful geological information for TBM drilling and tunnel construction. [ABSTRACT FROM AUTHOR]

Accurately predicting tunnel boring machine (TBM) performance is beneficial for excavation efficiency enhancement and risk mitigation of TBM tunneling. In this paper, we develop a long short-term memory (LSTM) based hybrid intelligent model to predict two key TBM performance parameters (advance rate and cutterhead torque). The model combines the LSTM, BN, Dropout and Dense layers to process the raw data and improve the fitting quality. The features, including the ground formation properties, tunnel route curvature, tunnel location and TBM operational parameters, are divided into historical/real-time time-varying parameters, time-invariant parameters and historical/real-time output prediction data. The effectiveness of the proposed model is verified based on a large monitoring database of the Baimang River Tunnel Project in Shenzhen, south China. We then discuss the influence of the prediction mode, neural network structure and time division interval length of historical data on the prediction accuracy. The significance evaluation of input features shows that the historical output prediction has the largest influence on the prediction accuracy, and the influence of ground properties is secondary. It is also found that the correlations between input features and the output prediction are coincident with their interrelationships with the ground properties and ease of TBM excavation. Finally, it is found that the prediction results are most affected by the total propulsion force followed by the rotation speed of the cutterhead. The established model can provide useful guidance for construction personnel to roughly grasp the possible TBM status from the prediction results when adjusting the operational parameters. [ABSTRACT FROM AUTHOR]

Detecting and analyzing small earthquakes is important for many seismological studies. Signals of small earthquakes are often obscured by noise. Recent advances in signal processing and deep learning along with available computing resources provide a great opportunity to address this challenge. In this study, we present a time domain method of suppressing noise for processing livestream earthquake data from a large seismic network by applying a deep neural network Real-time Denoiser (RTDenoiser). This neural network is able to attenuate a variety of colored noise and non-earthquake signals and suppress noise in the overlapping frequency bandwidth with signals. Because of its simplicity in time domain without domain transformation and subsequent processing, the method is able to process continuous livestream three-component data from several hundreds of seismic stations simultaneously. We create 'noise-free' samples by scaling down the waveforms of relatively large events from ML 3.5–5.0 to ML 1.5–3.0 according to the Richter scaling relationship. We also select noise samples from the same seismic station and add to 'noise-free' data to generate samples at different signal-to-noise ratio (SNR) levels. These data samples are randomly split into training, validation, and test sets. We verify the trained network to process data recorded in Sichuan and Yunnan, China from 2013 to 2018. Results show that the RTDenoiser can help improve SNRs from 5 dB to 15 dB in averag. The number of detected small events at magnitude between ML 1.0 and 3.0 has been increased by 58.8 per cent. The method is currently applied in a seismic network of 300 stations in Sichuan and Yunnan, China for continuous processing. It takes about 10 ms on average to process three-component 60-s data from 300 seismic stations on a single GPU. [ABSTRACT FROM AUTHOR]

Deep learning (DL) methods have a high potential for earthquake detection applications because of their high efficiency at processing measurement data, such as picking seismic phases. However, the performance of DL methods must be evaluated to ensure that they can replace conventional methods so that full automation can be achieved. State-of-art DL methods incorporate advanced techniques and train with large global datasets to enhance their earthquake detection capabilities. In this study, we tested a representative DL model on the 2016 Gyeongju earthquake sequence in the Korean Peninsula and compared the results with a previously established catalog and with the results of the conventional Short Time Average/Long Time Average (STA/LTA) method. The DL model demonstrated reasonable improvements in efficiency and performance by detecting more and smaller earthquakes within a much shorter running time than the other methods. In addition, the DL algorithms generally provided precise pickings of P- and S-wave phases. The DL model showed good generalization because it appropriately detected earthquakes in the study area that were not included in the training dataset. However, our results did suggest possible errors that should be accounted for, such as inconsistent phase picking, missing large earthquakes, and detecting non-natural earthquake signals. From the result of tests, local optimization may be important for realizing fully automatic earthquake monitoring, such as retraining with a local dataset, fine-tuning, or transfer learning. In addition, incorporating post-processing techniques such as phase association and discrimination into the DL framework is necessary. [ABSTRACT FROM AUTHOR]

Gridded sea surface height (SSH) maps estimated from satellite altimetry are widely used for estimating surface ocean geostrophic currents. Satellite altimeters observe SSH along one‐dimensional tracks widely spaced in space and time, making accurately reconstructing the two‐dimensional (2D) SSH field challenging. Traditionally, SSH is mapped using optimal interpolation (OI). However, OI artificially smooths the SSH field leading to high mapping errors in regions with rapidly‐evolving mesoscale features such as western boundary currents. Motivated by the dynamical relation between SSH and sea surface temperature (SST) and the notion that even the chaotic evolution of mesoscale ocean turbulence may contain repeating patterns, we outline a deep learning (DL) approach where a neural network is trained to reconstruct 2D SSH by synthesizing altimetry and SST observations. In the Gulf Stream Extension region, dominated by mesoscale variability, our DL method substantially improves the SSH reconstruction compared to existing methods. Our SSH map has 17% lower root‐mean‐square error and resolves spatial scales 30% smaller than OI compared against independent altimeter observations. Surface geostrophic currents calculated from our map are closer to surface drifter observations and appear qualitatively more realistic, with stronger currents, a clearer separation between the Gulf Stream and neighboring eddies, and the appearance of smaller coherent eddies missed by other methods. Our map yields significant re‐estimations of important dynamical quantities such as eddy kinetic energy, vorticity, and strain rate. Applying our DL method to produce a global SSH product may provide a more accurate and higher resolution product for studying mesoscale ocean turbulence. Plain Language Summary: Satellites observe small variations in the height of the sea surface but with large gaps in the observations. Having an estimate of the two‐dimensional sea surface height field allows one to estimate surface ocean currents, so filling in the gaps between the observations is an important problem. The traditionally‐used method for filling in the gaps between sea surface height observations struggles when there are lots of small‐scale, rapidly‐interacting ocean currents. We developed a deep learning (DL) model to estimate the sea surface height field more accurately. We achieved this by combining the sea surface height observations with satellite observations of sea surface temperature. The relationship between sea surface temperature and height is non‐trivial, but our DL model learned to use information from the sea surface temperature observations in the places where sea surface height wasn't observed to improve the accuracy of the sea surface height estimate. We applied and tested our method in the Gulf Stream and demonstrated that our sea surface height map is more accurate than that from the traditional method. Our method provides a more accurate sea surface height map which could allow us in future to learn new lessons about small‐scale surface currents in the ocean. Key Points: We developed a deep learning method that significantly improves the accuracy and resolution of gridded sea surface height anomaliesThis data‐driven method takes advantage of combining sea surface temperature and altimetry observationsInferred surface geostrophic currents are quantitatively and qualitatively more realistic than those from existing sea surface height maps [ABSTRACT FROM AUTHOR]

The detection of deformation in Global Navigation Satellite System (GNSS) time series associated with (a)seismic events down to a low magnitude is still a challenging issue. The presence of a considerable amount of noise in the data makes it difficult to reveal patterns of small ground deformation. Traditional analyses and methodologies are able to effectively retrieve the deformation associated with medium to large magnitude events. However, the automatic detection and characterization of such events is still a complex task, because traditionally employed methods often separate the time series analysis from the source characterization. Here we propose a first end‐to‐end framework to characterize seismic sources using geodetic data by means of deep learning, which can be an efficient alternative to the traditional workflow, possibly overcoming its performance. We exploit three different geodetic data representations in order to leverage the intrinsic spatio‐temporal structure of the GNSS noise and the target signal associated with (slow) earthquake deformation. We employ time series, images, and image time series to account for the temporal, spatial, and spatio‐temporal domain, respectively. Thereafter, we design and develop a specific deep learning model for each dataset. We analyze the performance of the tested models both on synthetic and real data from North Japan, showing that image time series of geodetic deformation can be an effective data representation to embed the spatio‐temporal evolution, with the associated deep learning method outperforming the other two. Therefore, jointly accounting for the spatial and temporal evolution may be the key to effectively detect and characterize fast or slow earthquakes. Plain Language Summary: The continuous monitoring of ground displacement with Global Navigation Satellite System allowed, at the beginning of the 2000s, the discovery of slow earthquakes—a transient slow slippage of tectonic faults that releases stress without generating seismic waves. Nevertheless, the detection of small events is still a challenge, because they are hidden in the noise. Most of the methods which are traditionally employed are able to extract the deformation down to a certain signal‐to‐noise level. However, one can ask if deep learning can be a more efficient and powerful alternative. To this end, we address the problem by using deep learning, as it stands as a powerful way to automatize and possibly overcome traditional methods. We use and compare three data representations, that is time series, images, and image time series of deformation, which account for the temporal, spatial, and spatio‐temporal variability, respectively. We train our methods on synthetic data, since real datasets are still not enough to be effectively employed with deep learning, and we test on synthetic and real data as well, claiming that image time series and its associated deep learning model may be more effective toward the study of the slow deformation. Key Points: We develop deep learning approaches on synthetics mimicking the spatio‐temporal structure of static deformation and realistic Global Navigation Satellite System (GNSS) noiseWe design three deep learning models and we train and test them against three GNSS data representationsTransformers and image time series of deformation can effectively characterize small deformation patterns associated with the seismic source [ABSTRACT FROM AUTHOR]

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