Location Prediction Research Articles

Deep Learning Models for Anatomical Location Classification in Esophagogastroduodenoscopy Images and Videos: A Quantitative Evaluation with Clinical Data

Background/Objectives: During gastroscopy, accurately identifying the anatomical locations of the gastrointestinal tract is crucial for developing diagnostic aids, such as lesion localization and blind spot alerts. Methods: This study utilized a dataset of 31,403 still images from 1000 patients with normal findings to annotate the anatomical locations within the images and develop a classification model. The model was then applied to videos of 20 esophagogastroduodenoscopy procedures, where it was validated for real-time location prediction. To address instability of predictions caused by independent frame-by-frame assessment, we implemented a hard-voting-based post-processing algorithm that aggregates results from seven consecutive frames, improving the overall accuracy. Results: Among the tested models, InceptionV3 demonstrated superior performance for still images, achieving an F1 score of 79.79%, precision of 80.57%, and recall of 80.08%. For video data, the InceptionResNetV2 model performed best, achieving an F1 score of 61.37%, precision of 73.08%, and recall of 57.21%. These results indicate that the deep learning models not only achieved high accuracy in position recognition for still images but also performed well on video data. Additionally, the post-processing algorithm effectively stabilized the predictions, highlighting its potential for real-time endoscopic applications. Conclusions: This study demonstrates the feasibility of predicting the gastrointestinal tract locations during gastroscopy and suggests a promising path for the development of advanced diagnostic aids to assist clinicians. Furthermore, the location information generated by this model can be leveraged in future technologies, such as automated report generation and supporting follow-up examinations for patients.

Prediction of Commercial Street Location Based on Point of Interest (POI) Big Data and Machine Learning

Identifying optimal locations for sustainable commercial street development is crucial for driving economic growth and enhancing social vitality in cities. This study proposes a data-driven approach to predict potential sites for commercial streets in Foshan City, China, utilizing Points of Interest (POI) big data and machine learning techniques. Decision tree algorithms are employed to quantitatively assess and predict optimal locations at a fine-grained spatial resolution, dividing the study area into 9808 grid cells. The analysis identifies 2157 grid cells as potential sites for commercial street development, highlighting the significant influence of Medical Care, Shopping, and Recreation and Entertainment POIs on site selection. The study underscores the importance of considering population base, human activity patterns, and cultural elements in sustainable urban development. The main contributions include providing a novel decision-support method for data-driven and sustainable commercial street site selection and offering insights into the complex interplay between urban land use, human activities, and commercial development. The findings have important implications for urban planning and policy-making, showcasing the potential of data-driven approaches in guiding sustainable urban development and fostering vibrant commercial areas.

Sparse-Laplace hybrid graph manifold method for fluorescence molecular tomography

Objective.Fluorescence molecular tomography (FMT) holds promise for early tumor detection by mapping fluorescent agents in three dimensions non-invasively with low cost. However, since ill-posedness and ill-condition due to strong scattering effects in biotissues and limited measurable data, current FMT reconstruction is still up against unsatisfactory accuracy, including location prediction and morphological preservation.Approach.To strike the above challenges, we propose a novel Sparse-Laplace hybrid graph manifold (SLHGM) model. This model integrates a hybrid Laplace norm-based graph manifold learning term, facilitating a trade-off between sparsity and preservation of morphological features. To address the non-convexity of the hybrid objective function, a fixed-point equation is designed, which employs two successive resolvent operators and a forward operator to find a converged solution.Main results.Through numerical simulations andin vivoexperiments, we demonstrate that the SLHGM model achieves an improved performance in providing accurate spatial localization while preserving morphological details.Significance.Our findings suggest that the SLHGM model has the potential to advance the application of FMT in biological research, not only in simulation but also inin vivostudies.

IFKC and Optimized DeTrac Deep Convolutional Neural Network Based Location Prediction in the Mobile IOT

Wireless Sensor Network (WSN) is a noteworthy technological advancement in the Internet of Things (IoT). Due to its severe resource constraints, this network necessitates the development of energy-efficient routing strategies. In the context of the Internet of Things, a cluster-based routing problems frequently struggle with unequal power usage. Therefore, this paper proposes an Improved Fractional Rough Fuzzy K-means Clustering and Optimized DeTraC Deep Convolutional Neural Network based Location Prediction in Mobile Internet of Things (IFKC-ODTN-LP-MIoT) system to address unequal power consumption in cluster-based IoT routing. By proposing Improved Fractional Rough Fuzzy K-means (IF-RFKM) for cluster head selection and an optimized DeTraC Deep Convolutional Neural Network (DT-DCNN) the system aims to enhance energy efficiency for location prediction. Then Adolescent Identity Search Algorithm (AISA) optimizes DT-DCNN parameters for improved accuracy. The experimental results demonstrate significant improvements in Computational efficiency, network lifetime and delay compared to the existing approaches.

Bayesian optimized deep Q-network for diagnosing mine ventilation systems windage alteration fault targeting imbalanced data

Fault diagnosis of mine ventilation system is of great significance for mine safety production. Traditional machine learning algorithms have been widely applied in the field of mine ventilation systems windage alteration faults (WAFs) diagnosis, but these algorithms have poor intelligence and weak generalization ability. Therefore, this research constructs a WAFs diagnosis model (WAFs-BODQN) based on Bayesian optimized (BO) deep Q-network (DQN). Meanwhile, in order to solve the problem of poor performance of intelligent models when the dataset is imbalanced, a reinforcement learning reward function (KMeans-SDF) was designed, which combines the KMeans++ algorithm and spatial distance function (SDF). The WAFs-BODQN model with KMeans-SDF reward function has fault location diagnosis accuracies of 98.75 %, 97.50 %, and 95.00 % for mild, moderate, and extreme imbalanced datasets UB1, UB2, and UB3, respectively. This effectively avoids the "eccentricity" phenomenon of the intelligent model on majority class and achieves accurate prediction of fault locations. The accuracy of fault location diagnosis using the WAFs-BODQN model in simulation experiments and on-site empirical applications is higher than 95 %, demonstrating the intelligence, generalization, and feasibility of WAFs-BODQN model in the field of WAFs diagnosis in mine ventilation systems. By comparing the results with traditional machine learning and deep learning fault diagnosis models, it has been proven that the WAFs-BODQN model has advantages in intelligence and generalization ability. It is expected to establish a universal architecture for diagnosing WAFs, providing theoretical guidance and technical support for achieving intelligent mine ventilation.

Development of a gunshot location detection system based on acoustic data

Determination of sound source location is a crucial problem, especially for the assault threat detection in crowded areas or military applications. The chief objective of this study is to design a microphone array to detect the location of explosion type sound sources, such as gunshot, which have significant amplitudes when compared to the background noise levels. Furthermore, the time arrival difference algorithm is utilized for the determination of time arrival differences and sound source location. Evidently, for stationary microphones that are positioned close to each other, determination of time arrival differences with high accuracy is a significant problem. Thus, a significant effort is given to enhance the accuracy of time arrival difference predictions via proper sampling. The design process of the microphone array is conducted computationally, where the sound data received by the microphones are generated. The time arrival differences are then calculated numerically, which are used for the prediction of sound source location. The number of microphones, relative positions of the microphones are determined accordingly via digital experiments. Finally, the designed microphone array is built and experiments are conducted. In conclusion, it is observed that the numerical model successfully predicts the sound source location of the computational model.

Spatial distribution of poultry farms using point pattern modelling: A method to address livestock environmental impacts and disease transmission risks.

The distribution of farm locations and sizes is paramount to characterize patterns of disease spread. With some regions undergoing rapid intensification of livestock production, resulting in increased clustering of farms in peri-urban areas, measuring changes in the spatial distribution of farms is crucial to design effective interventions. However, those data are not available in many countries, their generation being resource-intensive. Here, we develop a farm distribution model (FDM), which allows the prediction of locations and sizes of poultry farms in countries with scarce data. The model combines (i) a Log-Gaussian Cox process model to simulate the farm distribution as a spatial Poisson point process, and (ii) a random forest model to simulate farm sizes (i.e. the number of animals per farm). Spatial predictors were used to calibrate the FDM on intensive broiler and layer farm distributions in Bangladesh, Gujarat (Indian state) and Thailand. The FDM yielded realistic farm distributions in terms of spatial clustering, farm locations and sizes, while providing insights on the factors influencing these distributions. Finally, we illustrate the relevance of modelling realistic farm distributions in the context of epidemic spread by simulating pathogen transmission on an array of spatial distributions of farms. We found that farm distributions generated from the FDM yielded spreading patterns consistent with simulations using observed data, while random point patterns underestimated the probability of large outbreaks. Indeed, spatial clustering increases vulnerability to epidemics, highlighting the need to account for it in epidemiological modelling studies. As the FDM maintains a realistic distribution of farm location and sizes, its use to inform mathematical models of disease transmission is particularly relevant for regions where these data are not available.

Predictive modeling of flow characteristics in supersonic separators using machine learning

In the pursuit of cleaner energy sources, pre-processing fuel streams is crucial before distribution. Challenges arise during this process, leading to operational issues and diminished transmission efficiency. Among different technologies, supersonic separators are the ones that separate undesired components from the main gas stream more effectively. Precise forecasting of nozzle flow rate and shockwave location at a low cost can markedly enhance the efficiency of manufacturing and rating. In this research, the one-dimensional method has been developed to solve flow equations simultaneously by implementing equations of state. CFD has been used for the verification of the developed model, and the effect of different equations of state is investigated. Results show that the one-dimensional model significantly reduces the calculation time. Different EOSs result in up to 2.0 % and 3.5 % variation in the prediction of shock location and the mass flow rate, respectively.A machine learning methodology is employed to predict the location of shockwaves and the choked flow rate at an even lower cost. A dataset of 160 entries, generated using Peng Robinson's EOS, serves as the basis for this analysis. Prediction tasks are executed using Random Forest, k-nearest Neighbors, and Multilayer Perceptron models. While all three models demonstrate proficiency in data interpolation with R2 scores up to 0.9962 and 0.9951 for the shockwave location and mass flow rate, respectively, discerning their performance in extrapolation appears contingent on the specific case. In the case of extrapolation, Multilayer Perceptron and k-Nearest Neighbors show better match to one-dimensional results.

Relationship between optic nerve length and interoptic angle in the prediction of optic chiasm location

Introduction and objectivesThe sellar region is an area in the base of the skull that is among the most common sites for tumors of the central nervous system. Surgical interventions are currently performed via different routes. While the optic chiasm occupies its expected position in 70% of the population, it can deviate from this position. In such cases, surgery involving this region becomes more difficult as the known surgical routes are narrowed. Advance awareness of these variations can help surgeons to identify the optimal route for safe surgical intervention in the sellar region. By performing simple measurements of both the lengths of the optic nerves and the angle between them, a surgeon can predict the location of the chiasm. Materials and methodsTwenty specimens collected from autopsies performed at Bursa Forensic Medicine İnstitute were examined to determine the optic chiasm types and the relationships between the surrounding subchiasmal structures. ResultsAmong the 20 specimens, we found two prefixed (10%), 10 normo-fixed (50%), and eight postfixed chiasms (40%). The mean interoptic angle was 81.03 (±17.41)⁰. Prefixed chiasms had angles in the range 115.36°–124.76 ° (mean 120.06 [±6.65]⁰), normo-fixed chiasm angles were between 83.11° and 97.53 ° (mean 86.07 [±6.73]⁰), and postfixed chiasms ranged between 53.01 ° and 78.71 ° (mean 69.20 [±9.13]⁰). The length of the right optic nerve ranged between 6.95 and 13.83 mm (mean 10.25 [±1.81] mm), and the length of the left between 7.25 and 12.51 mm (mean 10.40 [±1.47] mm). Obtuse angles indicated that the chiasm was prefixed, and acute angles were indicative of a postfixed chiasm. There was a strong negative correlation between optic nerve lengths and the interoptic angle; thus, as the length of the nerves increases, the interoptic angle becomes more acute. ConclusionsWe have proposed a simple measurement of the optic nerve lengths and the angle between them to predict the relative location of the OC, which can be done easily on MRI.

Quasi-static crushing response of a novel triaxial isotropy mechanical metamaterial with dual-platform property

A novel triaxial isotropy origami metamaterial with dual-platform is proposed by combining the tachi tubes and the honeycomb structures. Crushing responses of the hexahedral metamaterial under quasi-static compression load are investigated through experimental tests and numerical simulations. Experimental and numerical results reveal that the hexahedral metamaterial sample shows three deformation modes. Meanwhile, the numerical predictions of deformation modes and locations agree very well with the experimental results. Moreover, the effect of aspect ratio, thickness-to-span ratio, angle on the deformation mode, peak stress, plateau stress of different stages, and specific energy absorption (SEAM, SEAV) is investigated. Finally, the proposed metamaterial is compared with traditional honeycomb. The numerical results demonstrate that the SEAM of the hexahedral metamaterial is 90.6% of the traditional honeycomb in the Z-direction. However, during X/Y-direction compression, the energy absorption capacity of the hexahedral metamaterial is 13.0 and 12.2 times that of the traditional honeycomb, respectively.

Secure smart grid implementation with automatic data integrity attack location prediction and exalted energy theft detection

Smart grid systems use digital communication technology to efficiently monitor and manage electricity distribution. However, there is a need for improve the robustness of smart grid systems. Hence a novel “Secure Smart Grid Implementation with Automatic Data Integrity Attack Location Prediction and Exalted Energy Theft Detection” has been proposed. Current detection methods are not good at identifying data integrity attacks. So a novel Automatic Deterministic Attack Location Prediction Model detects data integrity attacks using a Stacked Watermarking BiGRU-based Deep Deterministic Q network, extracting system matrix and state time series variation. Moreover, Energy theft detection algorithms, which are intended to detect dishonest clients, provide erroneous findings because of linear interpolation processes and randomised near misses, which ignore important data points. Thus a novel Exalted Energy Theft Detector with Secure Encoding is introduced where Amortized Multicode GAN method secures smart grids by detecting energy theft and preventing firmware code modifications, utilising fractional multi-level quantum encoding to identify injection locations. As a result, the proposed model has an accuracy of 97.1%, recall of 96.25%, error rate is 1.36%, F1-score is 96.20%, sensitivity is 96.10%, specificity is 96.10%, MAPE is 0.48%, MSE is 0.0001 and RMSE is 0.017.

Electric Vehicle Next Charge Location Prediction

Electric Vehicle Next Charge Location Prediction

Geospatial Modelling of Flood Susceptibility of the Calabar City, Cross River State, Nigeria

Flooding remains a major environmental problem in many parts of the world including Nigeria, causing untold losses. This study modelled the flood susceptibility of Calabar Metropolis using frequency ratio and geographical information system (GIS). Direct field observation and key informant approaches were used to obtain 127 previous and current flood locations in the study area from which a database of historical flood occurrence was developed. Using the geostatistical analyst tool within a geographical information environment the flood locations were split into two parts in the ratio of 70:30 percent for the training and testing processes, respectively. Eight flood conditioning factors (elevation, slope, aspect, curvature topographic position index, topographic wetness index, land cover and normalized difference vegetation index) were extracted from SRTM-30m DEM and Landsat 8 data accordingly and used in the geospatial analysis of flood occurrence within the GIS. The frequency ratio model was then developed using the training dataset. The final product was the flood susceptibility map of the study area. The results revealed that 9.3 percent of the study area was considered to have very high flood susceptibility, 19.97 percent was classified as high flood susceptibility while 31.79 percent was under moderate risk area. About 39 percent of the study area was classified at least under low flood susceptibility. The order of contribution of the conditioning factors to the model were topographic position index (TPI), elevation, normalized difference vegetation index (NDVI), land cover, curvature, aspect, slope, and topographic wetness index (TWI). The validation of the model using the receiver operating characteristics (ROC) curve returned an AUC value of 74.81 percent. Hence, the result was acceptable for the prediction of flood locations in the study area. The findings would assist urban planners, environmental managers, risk reduction and management agencies, land developers, policy makers, and indeed the general public.

Potential nutrient, carbon and fisheries impacts of large-scale seaweed and shellfish aquaculture in Europe evaluated using operational oceanographic model outputs

Large-scale seaweed and shellfish aquaculture are increasingly being considered by policymakers as a source of food, animal feed and bioproducts for Europe. These aquacultures are generally thought to be low impact or even beneficial for marine ecosystems as they are ‘extractive’ – i.e., growing passively on foodstuff already available in seawater, and with potential habitat provision, fisheries, climate mitigation and eutrophication mitigation benefits. At some scale however, over-extraction of nutrients or chlorophyll could potentially have a negative effect on natural systems. Understanding the likely impacts of aquaculture production at scale is important to identify when safe limits are being approached. Taking seaweed aquaculture as the primary focus, this work uses operational oceanographic model outputs to drive prognostic growth models to predict the likely optimal distribution of seaweed farms across European waters to meet different production scenarios. A novel nutrient transport scheme is then used to model the interacting ‘footprints’ of nutrient drawdown from aquaculture facilities to demonstrate the likely spatial impact of large-scale aquaculture. Evaluation of both seaweed and shellfish contributions to CO2 balance under large scale production, and the potential impact on fisheries are also considered. The study finds that the impact of intensive seaweed aquaculture on nutrient availability could be significant where many farms are placed close together; but at the regional/basin scale even the highest level of production considered does not significantly impact total nutrient budgets. Seaweed aquaculture has the potential to extract large amounts of carbon dioxide, but the impact on carbon budgets depends on the end-use of the extracted seaweed. Shellfish aquaculture is a net source of CO2 due to the impact of calcification of shells on the carbonate system (i.e., alkalinity removal). However, gram-for-gram the CO2 impact of shellfish production is likely to be less than the impact of land-based meat production. Whilst operational oceanographic models are useful for taking a ‘broad brush’ approach to likely placement and impacts of aquaculture, reliable yield predictions for individual locations across European waters would require models integrating more physical and biogeochemical factors (wave environment, local currents, riverine inputs) at a finer scale than currently achievable.

A multi-point leakage prediction statistical method using Bayesian inference in water distribution networks

ABSTRACT Leakage in water distribution networks (WDNs) not only leads to serious water loss and pipe contamination but also affects residents’ daily water. Accurate localization of leaks in WDN is significant to conserve water resources and reduce economic losses. However, in traditional optimization and verification methods for leak detection, factors such as modeling and pressure monitoring point errors are not taken into account, resulting in the deviation from the actual location of leaks. To address the mentioned issues, this article presents a WDN leakage probability prediction method based on Bayesian inference. The method converts the prior probability of leakage events into posterior probability. Then, by utilizing the posterior probability density function, the uncertainty of modeling and measurement errors in the hydraulic simulation model are quantified. The probability distribution of leakage pipes and leakage quantity within the leakage area is calculated, allowing for the location prediction and corresponding magnitude of leakage. The experimental results indicate that the prediction model can detect unknown quantities of leakage events. By collecting multiple sets of leakage data, it is possible to accurately predict the location and quantity of leaks, enhancing the efficiency of leakage detection in large-scale water supply networks and providing decision-making assistance for water utilities.

Unveiling user interests: A deep user interest exploration network for sequential location recommendation

Sequential location recommendation, also known as next location prediction, is a crucial task aiming to compute the likelihood of a user visiting a certain point of interest (POI). It has significant applications in route planning and location-based advertisements. Despite the utilization of geographical information, existing methods often adopt a rigidly sequential prediction mode, focusing solely on predicting the next destination (i.e., where to go), which limits their ability to effectively model user trajectories in terms of predicting from where the user came. Moreover, these methods struggle to capture the evolving and deep user interests along their trajectory. To deal with these problems, we propose a novel approach called Deep User Interest Exploration Network (DUIEN) for sequential location recommendation. DUIEN incorporates two essential components to enhance recommendation accuracy. First, we employ a bidirectional neural network with the Cloze objective, enabling it to predict random masked POIs within user trajectories. This approach aids the model in gaining a better understanding of the context and dynamics of user trajectories. Second, we introduce a user interest exploration layer that seamlessly integrates into the sequential learning process. This layer effectively utilizes tag information associated with POIs and learns the intricate interplay between user interests and their trajectories using a deep user interest exploration method. In our experiments, DUIEN outperforms state-of-the-art sequential location recommendation models, demonstrating the effectiveness of the designed update mechanisms for deep user interest exploration. This showcases the potential of our approach to significantly enhance the accuracy of sequential location recommendation.

Adaptive formulation for probabilistic storm surge predictions through sharing of numerical simulation results across storm advisories

As a tropical storm/cyclone approaches landfall, real-time probabilistic predictions for the anticipated surge provide valuable information for emergency preparedness/response decisions. These probabilistic predictions are made through an uncertainty quantification process that involves: (i) generating a sufficiently large ensemble of storm scenarios based on the nominal storm advisory and the anticipated forecast errors; (ii) performing high-fidelity numerical simulations to obtain surge predictions for each storm scenario; and (iii) estimating surge statistics of interest by assembling the simulation results. This process is repeated whenever the nominal storm advisory is updated. The number of storm scenarios utilized in the analysis directly impacts the statistical accuracy of the probabilistic predictions; a larger ensemble improves accuracy but requires greater computational resources to provide predictions with the desired expediency to guide real-time decisions. This paper revisits two recently proposed Monte-Carlo (MC) frameworks that aim to improve accuracy without increasing the computational burden: adaptive importance sampling (AIS) and adaptive multi-fidelity Monte Carlo (AMFMC). The foundational concept behind them is similar: share numerical simulation results across the probabilistic predictions performed for different storm advisories to accelerate the MC estimation. This is achieved differently for each approach, through adaptive development of an importance sampling proposal density (for AIS) or a surrogate model (for AMFMC). Here, a direct comparison between these frameworks is established, focusing on the mechanisms for the information sharing and the challenges encountered in tuning the algorithm adaptive characteristics to provide probabilistic estimates across a large number of quantities of interest (QoIs), corresponding to the surge predictions for different locations within the coastal region of interest. As this large number results in conflicting choices for the adaptive characteristics, a compromise solution needs to be promoted. The efficacy of the two frameworks is examined in detail in this setting, comparing the accuracy of idealized implementations (adaptive decisions independently made for each QoI) to the accuracy of practical implementations (single, compromise decision within the MC implementation). The study also showcases the importance of information sharing across storm advisories in real-time probabilistic storm surge predictions and provides guidelines for an efficient adaptive MC formulation in such settings.

Estimation of top depth to underground targets of Karous-Hjelt and Fraser filtering of VLF-EM measurements: The Thumb's rule approach

Prediction of the exact location and depth of underground targets with the very low-frequency electromagnetic (VLF-EM) technique is one of the most important and difficult tasks in geophysical investigations. This study examined and compared the conventional 2D KIFFILT inversion pseudo-section and the use of Thumb's rule technique in the Fraser filter plot to estimate the top depth of underground targets. The VLF-EM measurement was performed over several empirical buried target models to identify anomalies or geophysical responses corresponding to subsurface targets. The Karous-Hjelt and Fraser filtering techniques were applied to estimate the depth of the identified anomalies using Thumb's rule and the conventional 2D KIFFILT inversion. The signal behaviours of the VLF-EM current density pseudo-sections and the application of Thumb's rule effectiveness in delineating empirical buried target models were examined. Thumb's rule shows 65 % accuracy with the actual depth of the empirical buried target models, while the conventional 2D KHFFILT inversion shows 30 % accuracy with the actual depth of burial. Thumb's rule is more effective and precise in predicting the accurate depth of underground targets. The influences of conductive and resistive materials on VLF-EM signals and the challenges of VLF-EM surveys were discussed. Thumb's rule is suggested as a substantial technique for estimating top depth to the underground target where depth estimation is of prime interest due to its large degree of accuracy. In addition, the total depth of the current density distribution was noted to be increased when the distance between measuring points increased. This means that VLF-EM signals with longer wavelengths indicate deeper depth penetration into the ground than signals with shorter wavelengths. The accuracy of Thumb's rule regarding top depth estimation of the anomalies has been successfully tested and validated, which can be used for VLF-EM investigation where accurate depth estimation is required. The VLF-EM technique can be considered reliable for depth estimation using Thumb's rule approach, which applies to a wide range of subsurface investigations.

Machine Learning‐Assisted Microearthquake Location Workflow for Monitoring the Newberry Enhanced Geothermal System

AbstractEnhanced geothermal systems (EGS) offer a sustainable energy source but face challenges in accurately locating microearthquakes induced during reservoir stimulation. Locating these microearthquakes provides reliable feedback on the stimulation progress. Current deep learning methods for locating earthquakes require extensive data sets for training, which is problematic as detected microearthquakes are often limited. To address the scarcity of training data, we propose a practical workflow using probabilistic multilayer perceptron (PMLP) which predicts microearthquake locations from cross‐correlation time lags in waveforms. Utilizing a 3D velocity model of Newberry site derived from ambient noise interferometry, we generate numerous synthetic microearthquakes and 3D acoustic waveforms for PMLP training. Accurate synthetic tests prompt us to apply the trained network to the 2012 and 2014 stimulation field waveforms. To enhance the accuracy of source localization, we carefully handpick the P‐arrival times. Predictions on the 2012 stimulation data set show major microseismic activity at depths of 0.5–1.2 km, correlating with a known casing leakage scenario. In the 2014 data set, the majority of predictions concentrate at 2.0–2.9 km depths, consistent with results obtained from conventional physics‐based inversion, and align with the presence of natural fractures from 2.0 to 2.7 km. We validate our findings by comparing the synthetic and field picks, demonstrating a satisfactory match for the first arrivals. By combining the benefits of quick inference speeds and accurate location predictions, we demonstrate the feasibility of using realistic synthetic data set to locate microseismicity for EGS monitoring.

Combining the spatiotemporal mobility patterns and MMC for next location prediction of fake base stations

The spatiotemporal mobility patterns and next location prediction of fake base stations (FBS) provide important technical support for the police to prevent spam messages from FBS. However, due to the difficulty in locating their real-time locations, our understanding of the mobility patterns and predictability of FBS is still limited. Based on the crowdsourced spam data, we extract the time and potential locations of FBS and propose a Tucker-MMC method that combines Tucker decomposition with a Mobility Markov Chain (MMC) model to investigate the mobility patterns and predictability of FBS sending spam messages. First, we utilize Tucker decomposition to reflect the spatial and temporal preferences during the movement of the corresponding FBS. Then the mobility regularity and the theoretical maximum predictability of the FBS trajectories with similar mobility preferences are analyzed by entropy and Fano's inequality. A Tucker-MMC is also established for the next location prediction. The results using the spam dataset in Beijing show that the accuracy of Tucker-MMC is more than double that of the MMC. The accuracy of the actual location prediction model is more likely to approach the theoretical maximum predictability when FBS send spam messages in a shorter time, shorter transfer distance, and smaller access range.