Location Prediction Research Articles

That's not the Mona Lisa! How to interpret spatial capture-recapture density surface estimates.

Spatial capture-recapture methods are often used to produce density surfaces, and these surfaces are often misinterpreted. In particular, spatial change in density is confused with spatial change in uncertainty about density. We illustrate correct and incorrect inference visually by treating a grayscale image of the Mona Lisa as an activity center intensity or density surface and simulating spatial capture-recapture survey data from it. Inferences can be drawn about the intensity of the point process generating activity centers, and about the likely locations of activity centers associated with the capture histories obtained from a single survey of a single realization of this process. We show that treating probabilistic predictions of activity center locations as estimates of the intensity of the process results in invalid and misleading ecological inferences, and that predictions are highly dependent on where the detectors are placed and how much survey effort is used. Estimates of the activity center density surface should be obtained by estimating the intensity of a point process model for activity centers. Practitioners should state explicitly whether they are estimating the intensity or making predictions of activity center location, and predictions of activity center locations should not be confused with estimates of the intensity.

IeMTLF: Interaction-enhanced Multi-Task Learning Framework for next location prediction

Location-Based Services (LBSs) provide spatial and semantic information of various locations, thanks to the advances in mobile devices and commercial map services. However, most existing studies on next location prediction neglect the semantic information and do not investigate the effects of different levels of semantic granularity. To address these issues, we propose an Interaction-enhanced Multi-Task Learning Framework (IeMTLF) that jointly predicts the next location and its semantic category and leverages higher-order feature interactions to enrich the semantic learning process. Moreover, we examine the effect of two different levels of semantic granularity, namely, specific and abstract, on prediction performance. We conducted extensive experiments on two real-world datasets and demonstrate that IeMTLF outperforms seven baselines on all evaluation metrics, with a maximum gain of more than 13% for Acc@5 compared to the state-of-the-art method. Our implementation code is available at https://github.com/mrc-wyh/IeMTLF.

Charging non-deterministic mobile nodes in a transfer learning approach

Wireless rechargeable sensor networks (WRSNs) provide a solution to the energy problem in wireless sensor networks by introducing chargers to recharge the sensor nodes with rechargeable batteries. Most of the existing studies on WRSN focus on charging static nodes or nodes with certain mobility, while a few works investigate charging non-deterministic mobile nodes, where the movement pattern of nodes is unknown. The main challenge in the study of charging non-deterministic nodes is how to find the energy-hungry nodes, because the movement of the nodes is uncertain. The existing schemes assume that the historical trajectories of each node are known and predict the future locations of the nodes by analyzing the trajectory data. However, these schemes require a large amount of trajectory data to train the prediction model, and it takes a considerable amount of time for the network to run to obtain this trajectory data. If there is a node energy depletion during this time, these schemes cannot perform the energy replenishment to the nodes. To address this problem, we propose a novel charging scheme to provide charging services for non-deterministic nodes by introducing the transfer learning technology to predict the future locations of non-deterministic nodes. In the proposed scheme, a backbone Graph Convolutional Network (GCN) is pre-trained with other trajectory datasets, based on which only a small amount of trajectory data of nodes in the target network is needed to complete the location prediction task. In addition, a new scheduling algorithm is proposed, which considers multiple states of the nodes, including the energy consumption rate, the remaining energy state and the future locations, to select the charging target nodes. Simulation results show that the proposed scheme is able to achieve energy replenishment of non-deterministic mobile nodes when the network is just starting to operate.

Exploring the Performance of an Artificial Intelligence-Based Load Sensor for Total Knee Replacements.

Using tibial sensors in total knee replacements (TKRs) can enhance patient outcomes and reduce early revision surgeries, benefitting hospitals, the National Health Services (NHS), stakeholders, biomedical companies, surgeons, and patients. Having a sensor that is accurate, precise (over the whole surface), and includes a wide range of loads is important to the success of joint force tracking. This research aims to investigate the accuracy of a novel intraoperative load sensor for use in TKRs. This research used a self-developed load sensor and artificial intelligence (AI). The sensor is compatible with Zimmer's Persona Knee System and adaptable to other knee systems. Accuracy and precision were assessed, comparing medial/lateral compartments inside/outside the sensing area and below/within the training load range. Five points were tested on both sides (medial and lateral), inside and outside of the sensing region, and with a range of loads. The average accuracy of the sensor was 83.41% and 84.63% for the load and location predictions, respectively. The highest accuracy, 99.20%, was recorded from inside the sensing area within the training load values, suggesting that expanding the training load range could enhance overall accuracy. The main outcomes were that (1) the load and location predictions were similar in accuracy and precision (p > 0.05) in both compartments, (2) the accuracy and precision of both predictions inside versus outside of the triangular sensing area were comparable (p > 0.05), and (3) there was a significant difference in the accuracy of load and location predictions (p < 0.05) when the load applied was below the training loading range. The intraoperative load sensor demonstrated good accuracy and precision over the whole surface and over a wide range of load values. Minor improvements to the software could greatly improve the results of the sensor. Having a reliable and robust sensor could greatly improve advancements in all joint surgeries.

Predicting the locations of missing persons in China by using NGO data and deep learning techniques

ABSTRACT Missing person crimes can seriously affect the well-being of Chinese families, and missing person destination prediction can help to solve this problem. Using nongovernmental organization (NGO) data to predict the locations of missing persons by random forest (RF) model has made progress. However, studies using these data have ignored the mass of oral information. Recent studies have demonstrated the effectiveness of oral information in detecting missing persons, but the impact on destination prediction remains unexplored. Therefore, this study proposes a missing person prediction (MP-Net) framework to incorporate oral information into missing person destination prediction and quantitatively describe the effect of different word properties on the prediction. The results show that compared to the baseline RF model, the proposed framework achieves a higher recall rate (87.18%) in the location prediction of missing persons. According to a quantitative word analysis, verbs and nouns in oral information significantly contributed to location prediction. After adjectives that might cause adverse effects were removed, the stability of the model was improved considerably. Overall, the findings of the proposed model and quantitative word analysis can help police or NGOs collect descriptive information in a targeted manner and make more accurate predictions about the whereabouts of missing persons.

Evaluating evidence for co-geography in the Anopheles-Plasmodium host-parasite system.

The often tight association between parasites and their hosts means that under certain scenarios, the evolutionary histories of the two species can become closely coupled both through time and across space. Using spatial genetic inference, we identify a potential signal of common dispersal patterns in the Anopheles gambiae and Plasmodium falciparum host-parasite system as seen through a between-species correlation of the differences between geographic sampling location and geographic location predicted from the genome. This correlation may be due to coupled dispersal dynamics between host and parasite but may also reflect statistical artifacts due to uneven spatial distribution of sampling locations. Using continuous-space population genetics simulations, we investigate the degree to which uneven distribution of sampling locations leads to bias in prediction of spatial location from genetic data and implement methods to counter this effect. We demonstrate that while algorithmic bias presents a problem in inference from spatio-genetic data, the correlation structure between A. gambiae and P. falciparum predictions cannot be attributed to spatial bias alone and is thus likely a genetic signal of co-dispersal in a host-parasite system.

GAS SOURCE LOCALIZATION THROUGH DEEP LEARNING METHOD BASED ON GAS DISTRIBUTION MAP DATABASE

The incident of harmful gas leakage can cause severe damage to the environment and several casualties to human beings while the gas localization system plays a major role in mitigating those causalities. With the advances in artificial intelligence technology, deep learning is able to enhance the accuracy of the gas localization system to locate the gas source. This paper proposes a gas localization system that utilizes three different deep learning models namely DNN, 1DCNN, and 2DCNN to locate the gas source within the gas map. The proposed method involves generating the gas distribution map through the large gas sensor array platform in real-world indoor scenarios. Those models are then trained using the collected database which allows for accurate prediction of the gas source location. The performance of each proposed deep learning model was compared to find the best model demonstrating the highest effectiveness in identifying gas leaks. The study has shown that the 1DCNN has the highest effectiveness in predicting the gas source in the range between 0.0 m to 0.3 m with 90.3% compared to the DNN and 2DCNN models.

A hybrid multiple input multiple output (MIMO) model for simultaneous localization and quantification of structural damage in beam structures

Minor cracks are common types of damage in beam-like structures that must be detected early to prevent complete structural failure. Detecting and addressing such damage is crucial for maintaining structural integrity, extending service life, and minimizing operational costs. In recent years, various methods for quantifying and localizing damage using vibration measurements have been developed. However, many of these methods struggle to quantify and accurately localize the damage simultaneously. This paper addresses this challenge by applying novel hybrid multi-input and multi-output (MIMO) regression algorithms. The proposed approach combines modal analysis data, which includes frequencies, with machine learning algorithms to achieve simultaneous damage localization and quantification in damaged beams. Both numerical and experimental modal analyses have been conducted to determine the first six modal frequencies. Multiple hybrid MIMO models have been trained and tested using the frequencies obtained. The trained models have been used to estimate damage location and severity, and the results are compared. The models’ performance was evaluated using seven key metrics, including mean squared error (MSE), coefficient of determination (R2), mean absolute error (MAE), and others. The Random Forest (RF) model demonstrated exceptional performance by achieving a significantly higher R2 value of 0.997 for the precise prediction of both damage location and magnitude. Optimizing these models through hyperparameter tuning significantly improved their accuracy and reliability. This approach exhibits superior efficacy, surpassing existing models, and holds promise for advancing structural health monitoring in critical applications.

Artificial Intelligence for 3D Reconstruction from 2D Panoramic X-rays to Assess Maxillary Impacted Canines.

The objective of this study was to explore the feasibility of current 3D reconstruction in assessing the position of maxillary impacted canines from 2D panoramic X-rays. A dataset was created using pre-treatment CBCT data from a total of 123 patients, comprising 74 patients with impacted canines and 49 patients without impacted canines. From all 74 subjects, we generated a dataset containing paired 2D panoramic X-rays and pseudo-3D images. This pseudo-3D image contained information about the location of the impacted canine in the buccal/lingual, mesial/distal, and apical/coronal positions. These data were utilized to train a deep-learning reconstruction algorithm, a generative AI. The location of the crown of the maxillary impacted canine was determined based on the output of the algorithm. The reconstruction was evaluated using the structure similarity index measure (SSIM) as a metric to indicate the quality of the reconstruction. The prediction of the impacted canine's location was assessed in both the mesiodistal and buccolingual directions. The reconstruction algorithm predicts the position of the impacted canine in the buccal, middle, or lingual position with 41% accuracy, while the mesial and distal positions are predicted with 55% accuracy. The mean SSIM for the output is 0.71, with a range of 0.63 to 0.84. Our study represents the first application of AI reconstruction output for multidisciplinary care involving orthodontists, periodontists, and maxillofacial surgeons in diagnosing and treating maxillary impacted canines. Further development of deep-learning algorithms is necessary to enhance the robustness of dental reconstruction applications.

Vehicle Trajectory Prediction Based on Local Dynamic Graph Spatiotemporal–Long Short-Term Memory Model

Traffic congestion and frequent traffic accidents have become the main problems affecting urban traffic. The effective location prediction of vehicle trajectory can help alleviate traffic congestion, reduce the occurrence of traffic accidents, and optimize the urban traffic system. Vehicle trajectory is closely related to the surrounding Point of Interest (POI). POI can be considered as the spatial feature and can be fused with trajectory points to improve prediction accuracy. A Local Dynamic Graph Spatiotemporal–Long Short-Term Memory (LDGST-LSTM) was proposed in this paper to extract and fuse the POI knowledge and realize next location prediction. POI semantic information was learned by constructing the traffic knowledge graph, and spatial and temporal features were extracted by combining the Graph Attention Network (GAT) and temporal attention mechanism. The effectiveness of LDGST-LSTM was verified on two datasets, including Chengdu taxi trajectory data in August 2014 and October 2018. The accuracy and robustness of the proposed model were significantly improved compared with the benchmark models. The effects of major components in the proposed model were also evaluated through an ablation experiment. Moreover, the weights of POI that influence location prediction were visualized to improve the interpretability of the proposed model.

Prediction of CHF location through applied machine learning

The prediction of critical heat flux (CHF) location is a challenging issue in power generation industries. Several factors viz. geometric parameters and operating conditions influence the occurrence of CHF, and many mechanisms have been proposed to better explain the physics behind the occurrence of CHF. As CHF is a complex phenomenon, its dependency on these factors exhibit non-linear relationship. Further, it is challenging to account for all the factors simultaneously in a single model. Moreover, the benchmark experimental data has to be available to validate the constructed model with all the factors viz., operating pressure and power, inlet mass flux and subcooling etc. Due to these reasons, the influence of these factors remain unresolved. Considering the above issues, the present study employs different machine learning models to predict CHF location and identify the accurate model based on the comparisons against the chosen benchmark experimental data. Becker et al. (1983) data bank was selected for machine learning model training as well as the testing due to the wide range of operating conditions covered. Hyperparameter tuning is applied to optimize the performance of the models. Among all the implemented models, the artificial neural network (ANN) model is found to perform well with the training and the test datasets. The mean absolute percentage error (MAPE) of the training and testing data points are 4.01% and 6.04%, respectively. The accuracy score of the ANN model is approximately 96% for the training and 94% for the testing datasets. The proposed model is helpful for design engineers and analysts of power plants for design optimization and to reduce the computational time.

Prediction of Traffic Incident Locations with a Geohash-Based Model Using Machine Learning Algorithms

This paper presents a novel geohash-based approach for predicting traffic incident locations using machine learning algorithms. The study utilized a three-stage model for predicting the locations of traffic incidents, which encompassed accidents, breakdowns, and other incidents. In the model, firstly, ArcGIS was used to convert the coordinates of traffic incidents into geohash areas, leading to the definition of incident locations. Secondly, variables affecting traffic incidents were extracted, and a dataset was created by utilizing the values of these variables in geohash fields. Finally, machine learning algorithms such as decision tree (DT), k-nearest neighbor (k-NN), random forest (RF), and support vector machine (SVM) algorithms were used to predict the geohash region of traffic incidents. After conducting hyperparameter optimization, we evaluated the efficacy of various machine learning algorithms in predicting the location of traffic incidents using different evaluation metrics. Our findings indicate that the RF, SVM, and DT models performed the best, with accuracy percentages of 91%, 88%, and 87%, respectively. The findings of the research revealed that traffic incident locations can be successfully predicted with the geohash-based forecasting model. The results offer traffic managers and emergency responders new perspectives on how to manage traffic incidents more effectively and improve drivers’ safety.

Prediction of JTE breakdown performance in SiC PiN diode radiation detectors using TCAD augmented machine learning

The design of Junction Termination Extension (JTE) is an important step for meeting the reliability requirement of SiC PiN diode radiation detectors. For early evaluation, Technology Computer Aided Design (TCAD) software is often used to simulate the electronic property of detectors with different JTE parameters. But it is time consuming, which need 1 h or even longer for one case. Here, a TCAD augmented Machine Learning (ML) method based on the fully connected Neural Network (NN) algorithm is proposed to predict the breakdown performance quickly with different parameters of Spatial Modulation (SM) JTE. Utilized ∼5000 datum generated by TCAD simulation, the ML model could be established and achieve good prediction of breakdown voltage and location within a few seconds. As a semi-supervised learning model, its prediction accuracy of breakdown location is higher than 89.4 % and the determination coefficient R2 of breakdown voltage could be up to 0.97 compared with TCAD simulations. Moreover, this model could give the relationship curve of breakdown voltages and doping concentration, which is useful to choose an ideal structure with a wide implantation dose window. Based on this ML prediction model, the design cycle of SiC PiN diode radiation detectors could be reduced significantly.

A light-weight edge-enabled knowledge distillation technique for next location prediction of multitude transportation means

In this article we study how we can transfer knowledge between mobility models that represent different locations and means of transport. Specifically, we propose the use of knowledge distillation and fine-tuning techniques in order to build accurate next location prediction models using a light-weight architecture that can significantly reduce the inference time.Our goal is not to add one more model in the mobility literature. Instead, we believe that it is of paramount importance to present how we can manage, specialize, and enhance well-trained mobility predictors. In addition to this, we take into consideration the ever-generating mobility data, the limited resources of the devices that run the models and we focus on how we can reduce their computational requirements. We have tried three variations on how we use knowledge distillation, namely distilled agent, double-distilled agent and pre-distilled agent with the latter having an overall improvement of 6.57% in the distance errors compared with a state-of-the-art next location prediction that does not use knowledge distillation and 99.8% reduction in inference time on edge devices with the utilization of light-weight Machine Learning frameworks such as, TensorFlow Lite.

Location Prediction Using Bayesian Optimization LSTM for RIS-Assisted Wireless Communications

Location Prediction Using Bayesian Optimization LSTM for RIS-Assisted Wireless Communications

Location prediction using forward geocoding for fire incident

Urban fires, although not a natural disaster, are a severe threat that often occurs in urban areas. Banjarmasin City, the capital of South Kalimantan Province and one of the most populous cities in Kalimantan, recorded 159 fire cases between 2020 and 2022, averaging nearly 53 cases yearly. In today’s digital era, people often share ongoing fire incidents using smartphones and update information on social media and online news. However, the resulting data could be more structured to serve as a dataset. This research addresses these issues by applying geocoding, a digital service that translates street addresses into geographic coordinates. This research uses three geocoders: Google Maps API, Bing Maps API, and Smart Monkey Geocoder. The accuracy of the three geocoders was tested using the Root Mean Square Error (RMSE) statistical method by comparing the geocoding results with valid locations. Prediction analysis was used to identify the next fire event through the density approach of the previous fire event points. This research is expected to provide insights into efficient data collection and structured data conversion, recommendations regarding the best geocoding service, and prediction of fire vulnerability locations based on recurring factors of fire incidents in the area. In conclusion, accurate data is the key to effective fire prediction.

Intelligent Fusion Framework for Predicting Defect Type and Localization in Steel Manufacturing Processes

The defect prediction in the manufacturing of steel is a critical challenge because it affects the quality and safety of the products. For this reason, intelligent image fusion approach is introduced in this research to enhance accurate prediction of defect types and locations in steel materials. By utilizing U-Net architecture and pretrained ResNet18 encoder layers, our method performs fusion of data from several imaging modalities thus supporting precise localization as well as classification of defects. In our model’s learning curves as well as comparing predicted segmentation masks with ground truth images, extensive experimentation and visualization show that our model captures subtle defects very well. By so doing, it exhibits robust performance that mitigates risks associated with overfitting since it can accurately identify any flaw while still having the ability to accept unseen data from other sources. These results suggest that our approach can highly contribute to improving quality control and safety standards for steel production.

Prediction of Transition Locations in Regions of the Minor Axis of Hypersonic Elliptic Cones using the BiGlobal-$e^N$ Method

Prediction of Transition Locations in Regions of the Minor Axis of Hypersonic Elliptic Cones using the BiGlobal-$e^N$ Method

QCD at finite temperature and density: Criticality

We overview recent theoretical developments in the search for QCD critical point at finite temperature and density, including from lattice QCD, effective QCD theories, and proton number cumulants in heavy-ion collisions. We summarize the available constraints and predictions for the critical point location and discuss future challenges and opportunities.

Prediction of Fuel Debris Location in Fukushima Nuclear Power Plant using Machine Learning

Accurate fuel debris location is crucial part of the decommissioning of the Fukushima Nuclear Power plants. Conventional methods face challenges due to extreme radiation and complex structure of the materials involved. In this study, we propose a novel approach utilising neutron detection and machine learning to estimate fuel material location. Geant4 simulations and pythonTM scripts have been used to generate a comprehensive dataset to train a machine learning model using MATLAB’s regression learner. A Gaussian Process Regression model was chosen for training and prediction. The results show excellent prediction performance to estimate the corium thickness effectively and to locate the nuclear fuel material with a mean square error (MSE) of 0.01. By combining the machine learning with nuclear simulation codes, this promises to enhance the nuclear decommissioning efforts to retrieve nuclear fuel debris.