Search

This study aims to improve the prediction accuracy of the computer-aided engineering (CAE) model for crashworthiness performance evaluation at speeds beyond those defined by current regulations and public domain testing protocols. One way of achieving this is by integrating data from a few physical crash tests with the CAE data using machine learning models. In this study, two scenarios are investigated: (1) improving CAE model prediction accuracy using test data of a vehicle type that is the same as that of the CAE model; (2) improving CAE model prediction accuracy using test data from two different types of vehicles (e.g., two different sizes of SUVs). In the first scenario, a novel approach is proposed in the displacement domain (deceleration versus displacement) to enable data fusion to help recover the unmodeled physics in the CAE model. A nonlinear spring-mass model is used to simulate rigid-barrier vehicle frontal impact. A Gaussian process regression (GPR) model is then applied in conjunction with a Gaussian mixture model to capture the model bias of the nonlinear spring constant under a dynamic analysis scheme. In the second scenario, we propose a time-domain method (deceleration versus time) based on temporal convolutional network (TCN) and transfer learning. An initial TCN model is first trained by fusing CAE data with physical test data of the first vehicle type based on data augmentation. This data-augmented TCN model is then fine-tuned through transfer learning using CAE and test data of the second vehicle type. It leverages the domain-invariant representations of the two types of vehicles to enhance the CAE model prediction accuracy of the second vehicle type. Case studies are used to validate the proposed approaches and to demonstrate their efficacy in improving the prediction accuracy of the CAE models. [ABSTRACT FROM AUTHOR]

Low-dimensional structure of data can solve the adversarial robustness-accuracy conflict for machine learning systems. Modern machine learning systems have demonstrated breakthrough performance in a multitude of applications. However, they are known to be highly vulnerable to small perturbations to the input data, known as adversarial attacks. There are many well-documented examples of such behavior, for example small perturbations of an image, which is imperceptible to a human, can significantly degrade performance of modern classifiers. Adversarial training has been put forward as a way to improve robustness of learning algorithms to adversarial attacks. However, this benefit often comes at the cost of decreasing accuracy on natural unperturbed inputs, pointing to a potential conflict between adversarial robustness and standard accuracy. In "Adversarial robustness for latent models: Revisiting the robust-standard accuracies tradeoff," Adel Javanmard and Mohammad Mehrabi develop a theory to show that when the data enjoys low-dimensional structure, then it is possible to train models that are nearly optimal with respect to both, the standard and robust accuracies. Over the past few years, several adversarial training methods have been proposed to improve the robustness of machine learning models against adversarial perturbations in the input. Despite remarkable progress in this regard, adversarial training is often observed to drop the standard test accuracy. This phenomenon has intrigued the research community to investigate the potential tradeoff between standard accuracy (a.k.a generalization) and robust accuracy (a.k.a robust generalization) as two performance measures. In this paper, we revisit this tradeoff for latent models and argue that this tradeoff is mitigated when the data enjoys a low-dimensional structure. In particular, we consider binary classification under two data generative models, namely Gaussian mixture model and generalized linear model, where the features data lie on a low-dimensional manifold. We develop a theory to show that the low-dimensional manifold structure allows one to obtain models that are nearly optimal with respect to both, the standard accuracy and the robust accuracy measures. We further corroborate our theory with several numerical experiments, including Mixture of Factor Analyzers (MFA) model trained on the MNIST data set. Funding: A. Javanmard was partially supported by the Sloan Research Fellowship in mathematics, an Adobe Data Science Faculty Research Award, the National Science Foundation Career Award DMS-1844481, and the National Science Foundation Award 2311024. Supplemental Material: The e-companion is available at https://doi.org/10.1287/opre.2022.0162. [ABSTRACT FROM AUTHOR]

Background: Choroid plexus (CP) volume has been recently proposed as a proxy for brain neuroinflammation in multiple sclerosis (MS). Purpose: To develop and validate a fast automatic method to segment CP using routinely acquired brain T1‐weighted and FLAIR MRI. Study Type: Retrospective. Population: Fifty‐five MS patients (33 relapsing–remitting, 22 progressive; mean age = 46.8 ± 10.2 years; 31 women) and 60 healthy controls (HC; mean age = 36.1 ± 12.6 years, 33 women). Field Strength/Sequence: 3D T2‐weighted FLAIR and 3D T1‐weighted gradient echo sequences at 3.0 T. Assessment: Brain tissues were segmented on T1‐weighted sequences and a Gaussian Mixture Model (GMM) was fitted to FLAIR image intensities obtained from the ventricle masks of the SIENAX. A second GMM was then applied on the thresholded and filtered ventricle mask. CP volumes were automatically determined and compared with those from manual segmentation by two raters (with 3 and 10 years' experience; reference standard). CP volumes from previously published automatic segmentation methods (freely available Freesurfer [FS] and FS‐GMM) were also compared with reference standard. Expanded Disability Status Scale (EDSS) score was assessed within 3 days of MRI. Computational time was assessed for each automatic technique and manual segmentation. Statistical Tests: Comparisons of CP volumes with reference standard were evaluated with Bland Altman analysis. Dice similarity coefficients (DSC) were computed to assess automatic CP segmentations. Volume differences between MS and HC for each method were assessed with t‐tests and correlations of CP volumes with EDSS were assessed with Pearson's correlation coefficients (R). A P value <0.05 was considered statistically significant. Results: Compared to manual segmentation, the proposed method had the highest segmentation accuracy (mean DSC = 0.65 ± 0.06) compared to FS (mean DSC = 0.37 ± 0.08) and FS‐GMM (0.58 ± 0.06). The percentage CP volume differences relative to manual segmentation were −0.1% ± 0.23, 4.6% ± 2.5, and −0.48% ± 2 for the proposed method, FS, and FS‐GMM, respectively. The Pearson's correlations between automatically obtained CP volumes and the manually obtained volumes were 0.70, 0.54, and 0.56 for the proposed method, FS, and FS‐GMM, respectively. A significant correlation between CP volume and EDSS was found for the proposed automatic pipeline (R = 0.2), for FS‐GMM (R = 0.3) and for manual segmentation (R = 0.4). Computational time for the proposed method (32 ± 2 minutes) was similar to the manual segmentation (20 ± 5 minutes) but <25% of the FS (120 ± 15 minutes) and FS‐GMM (125 ± 15 minutes) methods. Data Conclusion: This study developed an accurate and easily implementable method for automatic CP segmentation in MS using T1‐weighted and FLAIR MRI. Evidence Level: 1 Technical Efficacy: Stage 4 [ABSTRACT FROM AUTHOR]

The detection of two-line element (TLE) outliers and space events play a crucial role in enhancing spatial situational awareness. Therefore, this paper addresses the issue of TLE outlier detection methods that often overlook the mutual influence of multiple factors. Hence, a Multivariate Gaussian Mixture Model (MGMM) is introduced to consider the interdependencies among various indicators. Additionally, a Multi-strategy Genetic Algorithm (MGA) is employed to adjust the complexity of the MGMM, allowing it to accurately learn the actual distribution of TLE data. Initially, the proposed method applies probabilistic fits to the predicted error rate changes for both the TLE semi-major axis and the orbital inclination. Chaos initialization, a posterior probability penalty, and local optimization iterations are subsequently integrated into the genetic algorithm. These enhancements aim to estimate the MGMM parameters, addressing issues related to poor robustness and the susceptibility of the MGMM to converge to local optima. The algorithm's effectiveness is validated using TLE data from typical space targets. The results demonstrate that the optimized algorithm can efficiently detect outliers and maneuver events within complex TLE data. Notably, the comprehensive detection performance index, measured, using the F1 score, improved by 15.9% compared to the Gaussian mixture model. This significant improvement underscores the importance of the proposed method in bolstering the security of complex space environments. [ABSTRACT FROM AUTHOR]

Reinforced concrete slabs are widely used in building structures due to their economic, durable, and aesthetic advantages. The determination of their ultimate strength often hinges on punching shear strength. Presently, methods such as closed hoops, steel bending, and fiber reinforcement are employed to enhance punching shear strength, with fiber reinforcement gaining popularity due to its ease of implementation and efficacy in improving concrete durability. This study introduces a novel approach employing six machine learning algorithms rooted in decision trees and decision tree-based ensemble learning to predict punching shear strength in steel fiber-reinforced concrete slabs. To overcome experimental data limitations, a data augmentation approach based on the Gaussian mixture model is employed. The validation of the data augmentation is conducted through "synthetic training—real testing" and "real training—real testing". Additionally, the best machine learning model is analyzed for explainability using Shapley Additive exPlanation (SHAP). Results demonstrate that the proposed data augmentation method effectively captures the original data distribution, enhancing the robustness and accuracy of the machine learning model. Moreover, SHAP provides better insights into the features influencing punching shear strength. Thus, the proposed data enhancement model offers a reliable approach for modeling small experimental datasets in structural engineering. [ABSTRACT FROM AUTHOR]

Summary: Clustering ensemble is a popular approach for identifying data clusters that combines the clustering results from multiple base clustering algorithms to produce more accurate and robust data clusters. However, the performance of clustering ensemble algorithms is highly dependent on the quality of clustering members. To address this problem, this paper proposes a member enhancement‐based clustering ensemble (MECE) algorithm that selects the ensemble members by considering their distribution consistency. MECE has two main components, called heterocluster splitting and homocluster merging. The first component estimates two probability density functions (p.d.f.s) estimated on the sample points of an heterocluster and represents them using a Gaussian distribution and a Gaussian mixture model. If the random numbers generated by these two p.d.f.s have different probability distributions, the heterocluster is then split into smaller clusters. The second component merges the clusters that have high neighborhood densities into a homocluster, where the neighborhood density is measured using a novel evaluation criterion. In addition, a co‐association matrix is presented, which serves as a summary for the ensemble of diverse clusters. A series of experiments were conducted to evaluate the feasibility and effectiveness of the proposed ensemble member generation algorithm. Results show that the proposed MECE algorithm can select high quality ensemble members and as a result yield the better clusterings than six state‐of‐the‐art ensemble clustering algorithms, that is, cluster‐based similarity partitioning algorithm (CSPA), meta‐clustering algorithm (MCLA), hybrid bipartite graph formulation (HBGF), evidence accumulation clustering (EAC), locally weighted evidence accumulation (LWEA), and locally weighted graph partition (LWGP). Specifically, MECE algorithm has the nearly 23% higher average NMI, 27% higher average ARI, 15% higher average FMI, and 10% higher average purity than CSPA, MCLA, HBGF, EAC, LWEA, and LWGA algorithms. The experimental results demonstrate that MECE algorithm is a valid approach to deal with the clustering ensemble problems. [ABSTRACT FROM AUTHOR]

The underwater domain is a challenging environment for robotics because widely used electromagnetic devices must be substituted by acoustic equivalents, much slower and noisier. In this paper a two‐dimensional pose simultaneous localization and mapping (SLAM) system for an Autonomous Underwater Vehicle based on inertial sensors and a mechanical profiling sonar is presented. Two main systems are specially designed. On the one hand, a dead reckoning system based on Lie Theory is presented to track integrated pose uncertainty. On the other hand, a rigid scan matching technique specialized for acoustic data is proposed, which allows one to estimate the uncertainty of the matching result. Moreover, Bayesian–Gaussian mixtures models are introduced to the scan matching problem and the registration problem is solved by an optimization in Lie groups. The SLAM system is tested on real data and executed in real time with the robotic application. Using this system, section maps at constant depth can be obtained from a three‐dimensional underwater domain. The presented SLAM system constitutes the first achievement towards an underwater Active SLAM application. [ABSTRACT FROM AUTHOR]

The research area of the present study is the Sumatra megathrust zone, which can be partitioned into five segments based on the large earthquake sources, including the Aceh Andaman, Nias Simeulue, Mentawai Siberut, Mentawai Pagai, and Enggano segments. This work presents the recognition of seismicity patterns in the research area from January 1970 to December 2022 using segmental and zonal mathematical modeling of the annual maximum earthquake magnitude. To achieve this, we use two kinds of Gaussian mixture models: G-group Gaussian independent mixture models (G-group GMMs) and N-state Gaussian hidden Markov models (N-state GHMMs) to determine the appropriate probability density function of the seismicity data (ePDF). The fit model is selected based on the smallest Bayes information criterion. For the segment analysis, the results show that the ePDF of the Mentawai-Pagai segment fits the 2-state GHMM, whereas, for the four remaining segments, it tends to fit the 2-group GMM. Subsequently, for the zone analysis, the ePDF of the data fits the 2-state GHMM. Thus, from a segmental and zoning point of view, seismicity patterns fluctuate at two levels. From a seismic risk management aspect, these findings can be used to evaluate the risk vulnerability of an area to destructive earthquakes. That is, the patterns of seismicity sequences in all segments of the Sumatra megathrust zone all fluctuate within the range of moderate to strong earthquakes. Furthermore, the seismicity pattern in the Mentawai-Pagai segment and the Sumatra megathrust zone has Markov properties. [ABSTRACT FROM AUTHOR]

INTRODUCTION: With the continuous development of video analysis technology, sports video object tracking has become one of the research hotspots. Particle filter, as an effective object tracking algorithm, has been widely used in sports video object tracking. However, traditional particle filters have some shortcomings, such as particle depletion and high computational complexity, which require optimization. This article proposes a sports video object tracking algorithm based on optimized particle filters, aiming to improve the accuracy and stability of object tracking. Particle filter-based human motion video target tracking technology has become a trend. This project intends to apply particle filters to image processing of human activities. Firstly, an improved particle filter model tracks moving video objects. The purpose is to improve the tracking effect further and increase the accuracy. HSV distribution model was used to establish a target observation model. The algorithm is combined with the weight reduction algorithm to realize the human motion trajectory detection in the target observation mode. An examination of sports player videos then confirmed the model. Experiments show that this method can be used to track people in moving images of sports. Compared with other methods, this method has higher computational accuracy and speed. [ABSTRACT FROM AUTHOR]

In industrial environments, acoustic detection of equipment can often significantly change the acoustic characteristics between training and testing data due to the absence of anomalous data guidance, as well as changes in machine operating conditions or environmental noise. In addition, during model learning, there are cases where the target data is inaccessible, but an accurate model is still needed for the invisible target domain. To address these challenges, we propose an unsupervised approach for anomalous sound detection applied to unlabeled anomalous sound detection scenarios, which is an optimization algorithm based on joint deep learning and variational Gaussian mixture models, by two jointly training neural networks for feature extraction and using the variational Gaussian mixture model to analyze the obtained embeddings for clustering. We propose a new hybrid example data enhancement method, to generate examples in multiple ways, combining various methods to align the distribution between different domains. The improved sub-cluster AdaCos loss function is used to exclude potential anomalies by learning fewer finite distributions than the standard AdaCos. Experimental results show that the proposed method achieves an average area under curve of 82.70 % and an average F1 score of 69.42 % on the datasets of seven industrial equipment machine types. The average area under curve for the source domain is 85.49 % and for the target domain is 75.63 % . [ABSTRACT FROM AUTHOR]

Background: Genetic variants that severely alter protein products (e.g. nonsense, frameshift) are often associated with disease. For some genes, these predicted loss-of-function variants (pLoFs) are observed throughout the gene, whilst in others, they occur only at specific locations. We hypothesised that, for genes linked with monogenic diseases that display incomplete penetrance, pLoF variants present in apparently unaffected individuals may be limited to regions where pLoFs are tolerated. To test this, we investigated whether pLoF location could explain instances of incomplete penetrance of variants expected to be pathogenic for Mendelian conditions. Methods: We used exome sequence data in 454,773 individuals in the UK Biobank (UKB) to investigate the locations of pLoFs in a population cohort. We counted numbers of unique pLoF, missense, and synonymous variants in UKB in each quintile of the coding sequence (CDS) of all protein-coding genes and clustered the variants using Gaussian mixture models. We limited the analyses to genes with ≥ 5 variants of each type (16,473 genes). We compared the locations of pLoFs in UKB with all theoretically possible pLoFs in a transcript, and pathogenic pLoFs from ClinVar, and performed simulations to estimate the false-positive rate of non-uniformly distributed variants. Results: For most genes, all variant classes fell into clusters representing broadly uniform variant distributions, but genes in which haploinsufficiency causes developmental disorders were less likely to have uniform pLoF distribution than other genes (P < 2.2 × 10−6). We identified a number of genes, including ARID1B and GATA6, where pLoF variants in the first quarter of the CDS were rescued by the presence of an alternative translation start site and should not be reported as pathogenic. For other genes, such as ODC1, pLoFs were located approximately uniformly across the gene, but pathogenic pLoFs were clustered only at the end, consistent with a gain-of-function disease mechanism. Conclusions: Our results suggest the potential benefits of localised constraint metrics and that the location of pLoF variants should be considered when interpreting variants. [ABSTRACT FROM AUTHOR]

Background: We hypothesized that postural instability observed in individuals with Parkinson's disease (PD) can be classified as distinct subtypes based on comprehensive analyses of various evaluated parameters obtained from time-series of center of pressure (CoP) data during quiet standing. The aim of this study was to characterize the postural control patterns in PD patients by performing an exploratory factor analysis and subsequent cluster analysis using CoP time-series data during quiet standing. Methods: 127 PD patients, 47 aged 65 years or older healthy older adults, and 71 healthy young adults participated in this study. Subjects maintain quiet standing for 30 s on a force platform and 23 variables were calculated from the measured CoP time-series data. Exploratory factor analysis and cluster analysis with a Gaussian mixture model using factors were performed on each variable to classify subgroups based on differences in characteristics of postural instability in PD. Results: The factor analysis identified five factors (magnitude of sway, medio-lateral frequency, anterio-posterior frequency, component of high frequency, and closed-loop control). Based on the five extracted factors, six distinct subtypes were identified, which can be considered as subtypes of distinct manifestations of postural disorders in PD patients. Factor loading scores for the clinical classifications (younger, older, and PD severity) overlapped, but the cluster classification scores were clearly separated. Conclusions: The cluster categorization clearly identifies symptom-dependent differences in the characteristics of the CoP, suggesting that the detected clusters can be regarded as subtypes of distinct manifestations of postural disorders in patients with PD. [ABSTRACT FROM AUTHOR]

Reliable urban navigation using global navigation satellite system requires accurately estimating vehicle position despite measurement faults and monitoring the trustworthiness (or integrity) of the estimated location. However, reflected signals in urban areas introduce biases (or faults) in multiple measurements, while blocked signals reduce the number of available measurements, hindering robust localization and integrity monitoring. This paper presents a novel particle filter framework to address these challenges. First, a Bayesian fault-robust optimization task, formulated through a Gaussian mixture model (GMM) measurement likelihood, is integrated into the particle filter to mitigate faults in multiple measurement for enhanced positioning accuracy. Building on this, a novel test statistic leveraging the particle filter distribution and the GMM likelihood is devised to monitor the integrity of the localization by detecting errors exceeding a safe threshold. The performance of the proposed framework is demonstrated on real-world and simulated urban driving data. Our localization algorithm consistently achieves smaller positioning errors compared to existing filters under multiple faults. Furthermore, the proposed integrity monitor exhibits fewer missed and false alarms in detecting unsafe large localization errors. [ABSTRACT FROM AUTHOR]

Accurate measurement of abdominal aortic aneurysm is essential for selecting suitable stent-grafts to avoid complications of endovascular aneurysm repair. However, the conventional image-based measurements are inaccurate and time-consuming. We introduce the automated workflow including semantic segmentation with active learning (AL) and measurement using an application programming interface of computer-aided design. 300 patients underwent CT scans, and semantic segmentation for aorta, thrombus, calcification, and vessels was performed in 60–300 cases with AL across five stages using UNETR, SwinUNETR, and nnU-Net consisted of 2D, 3D U-Net, 2D-3D U-Net ensemble, and cascaded 3D U-Net. 7 clinical landmarks were automatically measured for 96 patients. In AL stage 5, 3D U-Net achieved the highest dice similarity coefficient (DSC) with statistically significant differences (p < 0.01) except from the 2D–3D U-Net ensemble and cascade 3D U-Net. SwinUNETR excelled in 95% Hausdorff distance (HD95) with significant differences (p < 0.01) except from UNETR and 3D U-Net. DSC of aorta and calcification were saturated at stage 1 and 4, whereas thrombus and vessels were continuously improved at stage 5. The segmentation time between the manual and AL-corrected segmentation using the best model (3D U-Net) was reduced to 9.51 ± 1.02, 2.09 ± 1.06, 1.07 ± 1.10, and 1.07 ± 0.97 min for the aorta, thrombus, calcification, and vessels, respectively (p < 0.001). All measurement and tortuosity ratio measured − 1.71 ± 6.53 mm and − 0.15 ± 0.25. We developed an automated workflow with semantic segmentation and measurement, demonstrating its efficiency compared to conventional methods. [ABSTRACT FROM AUTHOR]

Background: Using in vivo neuroimaging techniques, growing evidence has demonstrated that the choroid plexus (CP) volume is enlarged in patients with several neurodegenerative diseases, including Alzheimer's disease and Parkinson's disease. However, although animal and postmortem findings suggest that CP abnormalities are likely important pathological mechanisms underlying amyotrophic lateral sclerosis (ALS), the third most common neurodegenerative disease, no available study has been conducted to thoroughly assess CP abnormalities and their clinical relevance in vivo in ALS patients to date. Thus, we aimed to determine whether in vivo CP enlargement may occur in ALS patients. We also aimed to identify the relationships of CP volume with clinical disabilities and blood-CSF barrier (BCSFB) permeability in ALS patients. Methods: In this retrospective study, based on structural MRI data, CP volume was assessed using a Gaussian mixture model and underwent further manual correction in 155 ALS patients and 105 age- and sex-matched HCs from October 2021 to April 2023. The ALS Functional Rating Scale-Revised (ALSFRS-R) was used to assess clinical disability. The CSF/serum albumin quotient (Qalb) was used to assess BCSFB permeability. Moreover, all the ALS patients completed genetic testing, and according to genetic testing, the ALS patients were further divided into genetic ALS subgroup and sporadic ALS subgroup. Results: We found that compared with HCs, ALS patients had a significantly higher CP volume (p < 0.001). Moreover, compared with HCs, CP volume was significantly increased in both ALS patients with and without known genetic mutations after family-wise error correction (p = 0.006 and p < 0.001, respectively), while there were no significant differences between the two ALS groups. Furthermore, the CP volume was significantly correlated with the ALSFRS-r score (r = -0.226; p = 0.005) and the Qalb (r = 0.479; p < 0.001) in ALS patients. Conclusion: Our study first demonstrates CP enlargement in vivo in ALS patients, and continues to suggest an important pathogenetic role for CP abnormalities in ALS. Moreover, assessing CP volume is likely a noninvasive and easy-to-implement approach for screening BCSFB dysfunction in ALS patients. [ABSTRACT FROM AUTHOR]

With the greatly increased penetration rate of wind power, photovoltaic, and other new energy sources in the power system, the proportion of controllable units gradually decreased, resulting in increased system uncertainty. The biogas power generation system can effectively alleviate the pressure caused by source-load uncertainty in such high-permeability systems of new energy sources such as wind power and photovoltaic. Hence, from the perspective of the power system, this paper introduces a capacity demand analysis method for a rural biogas power generation system capable of independent operation amidst source-load uncertainty. To enhance the depiction of pure load demand uncertainty, a scene set generation method is proposed, leveraging quantile regression analysis and Gaussian mixture model clustering. Each scene's data and probability of occurrence elucidate the uncertainty of pure load demand. An integrated optimal operation model for new energy and biogas-generating units, free from energy storage capacity constraints, is established based on the generated scenario set. Addressing considerations such as biogas utilization rate and system operation cost, a biogas storage correction model, utilizing the gas storage deviation degree index and the cost growth rate index, is developed to determine biogas demand and capacity. The example results demonstrate the significant reduction in gas storage construction costs and charging and discharging imbalances achieved by the proposed model while ensuring systemic operational cost effectiveness. [ABSTRACT FROM AUTHOR]

This study aimed to use a data-driven approach to identify individualized speed thresholds to characterize running demands and athlete workload during games and practices in skill and linemen football players. Data were recorded from wearable sensors over 28 sessions from 30 male Canadian varsity football athletes, resulting in a total of 287 performances analyzed, including 137 games and 150 practices, using a global positioning system. Speed zones were identified for each performance by fitting a 5-dimensional Gaussian mixture model (GMM) corresponding to 5 running intensity zones from minimal (zone 1) to maximal (zone 5). Skill players had significantly higher (p < 0.001) speed thresholds, percentage of time spent, and distance covered in maximal intensity zones compared to linemen. The distance covered in game settings was significantly higher (p < 0.001) compared to practices. This study highlighted the use of individualized speed thresholds to determine running intensity and athlete workloads for American and Canadian football athletes, as well as compare running performances between practice and game scenarios. This approach can be used to monitor physical workload in athletes with respect to their tactical positions during practices and games, and to ensure that athletes are adequately trained to meet in-game physical demands. [ABSTRACT FROM AUTHOR]

This study aims to enhance diagnostic capabilities for optimising the performance of the anaerobic sewage treatment lagoon at Melbourne Water's Western Treatment Plant (WTP) through a novel machine learning (ML)-based monitoring strategy. This strategy employs ML to make accurate probabilistic predictions of biogas performance by leveraging diverse real-life operational and inspection sensor and other measurement data for asset management, decision making, and structural health monitoring (SHM). The paper commences with data analysis and preprocessing of complex irregular datasets to facilitate efficient learning in an artificial neural network. Subsequently, a Bayesian mixture density neural network model incorporating an attention-based mechanism in bidirectional long short-term memory (BiLSTM) was developed. This probabilistic approach uses a distribution output layer based on the Gaussian mixture model and Monte Carlo (MC) dropout technique in estimating data and model uncertainties, respectively. Furthermore, systematic hyperparameter optimisation revealed that the optimised model achieved a negative log-likelihood (NLL) of 0.074, significantly outperforming other configurations. It achieved an accuracy approximately 9 times greater than the average model performance (NLL = 0.753) and 22 times greater than the worst performing model (NLL = 1.677). Key factors influencing the model's accuracy, such as the input window size and the number of hidden units in the BiLSTM layer, were identified, while the number of neurons in the fully connected layer was found to have no significant impact on accuracy. Moreover, model calibration using the expected calibration error was performed to correct the model's predictive uncertainty. The findings suggest that the inherent data significantly contribute to the overall uncertainty of the model, highlighting the need for more high-quality data to enhance learning. This study lays the groundwork for applying ML in transforming high-value assets into intelligent structures and has broader implications for ML in asset management, SHM applications, and renewable energy sectors. [ABSTRACT FROM AUTHOR]

Background: Deposits of amyloid-β (Aβ) appear early in Alzheimer's disease (AD). Objective: The aim of the present study was to compare the presence of cortical and subcortical Aβ in early AD using positron emission tomography (PET). Methods: Eight cognitively unimpaired (CU) subjects, 8 with mild cognitive impairment (MCI) and 8 with mild AD were examined with PET and [11C]AZD2184. A data driven cut-point for Aβ positivity was defined by Gaussian mixture model of isocortex binding potential (BPND) values. Results: Sixteen subjects (3 CU, 5 MCI and 8 AD) were Aβ-positive. BPND was lower in subcortical and allocortical regions compared to isocortex. Fifteen of the 16 Aβ-positive subjects displayed Aβ binding in striatum, 14 in thalamus and 10 in allocortical regions. Conclusions: Aβ deposits appear to be widespread in early AD. It cannot be excluded that deposits appear simultaneously throughout the whole brain which has implications for improved diagnostics and disease monitoring. [ABSTRACT FROM AUTHOR]

Various published COVID-19 models have been used in epidemiological studies and healthcare planning to model and predict the spread of the disease and appropriately realign health measures and priorities given the resource limitations in the field of healthcare. However, a significant issue arises when these models need help identifying the distribution of the constituent variants of COVID-19 infections. The emergence of such a challenge means that, given limited healthcare resources, health planning would be ineffective and cost lives. This work presents a universal neural network (NN) computational instrument for predicting the mainstream symptomatic infection rate of COVID-19 and models of the distribution of its associated variants. The NN is based on a mixture density network (MDN) with a Gaussian mixture model (GMM) object as a backbone. Twelve use cases were used to demonstrate the validity and reliability of the proposed MDN. The use cases included COVID-19 data for Canada and Saudi Arabia, two date ranges (300 and 500 days), two input data modes, and three activation functions, each with different implementations of the batch size and epoch value. This array of scenarios provided an opportunity to investigate the impacts of epistemic uncertainty (EU) and aleatoric uncertainty (AU) on the prediction model's fitting. The model accuracy readings were in the high nineties based on a tolerance margin of 0.0125. The primary outcome of this work indicates that this easy-to-use universal MDN helps provide reliable predictions of COVID-19 variant distributions and the corresponding synthesized profile of the mainstream infection rate. [ABSTRACT FROM AUTHOR]

In this paper, the problem of state estimation for systems over wireless networks using user datagram protocol is focused on. It is known that for such a system, the probability density function of the system state follows a Gaussian mixture model (GMM), and the number of components in this model grows exponentially over time, which makes the computation of optimal estimates infeasible. To compute optimal estimates, based on Kullback‐Leibler divergence, a strategy with variable step‐sizes to truncate and fuse the GMM is proposed. Based on the obtained GMM, a variable step‐size estimator is designed to compute optimal estimates during an estimation cycle. The advantages of the proposed estimator are twofold: (1) its estimation performance is superior to that of existing one‐step fast estimators; (2) its estimation efficiency is much higher than that of the optimal estimator. Finally, trajectory tracking has been proposed for a real‐world hydrogen‐powered unmanned aerial vehicle to show the effectiveness of our methods. [ABSTRACT FROM AUTHOR]

The inaccuracy of inhomogeneous sound speed fields in photoacoustic imaging (PAI) can lead to the blurring and distortion of photoacoustic images. To solve this problem, conventional methods build speed models by using some a priori information or additional measuring equipment, which limits the application of PAI greatly. A data-driven speed field inversion method is proposed in this paper. It combines clustering with updates to the speed field. To reduce the complexity of the sound speed field model, the model is divided according to the similarity of the same tissue. The sound speed of the same tissue is regarded as a whole, which reduces the number of sound speed parameter solutions. Based on the simplified sound speed field model, the proposed method can adaptively adjust the step length of the sound speeds of various tissues by weight allocation. In this way, the updated amplitude of sound speeds of various tissues can be balanced and the accuracy of the sound speed field can be improved. A digital breast model is applied to verify the effectiveness of the proposed method. The results demonstrate that the method can build an appropriate speed field without additional information or equipment and improve the imaging performance of PAI. [ABSTRACT FROM AUTHOR]

The progress of Wireless Sensor Networks (WSNs) technologies has introduced a greater susceptibility of sensors and networks to being victims of distributed attacks. These attacks include various malicious activities such as intrusions during routing processes, data intercepting and other disruptive actions. In response to this increasing security challenge, numerous models for attack identification have been proposed. These models typically involve the deployment of detection systems that collect sensor data and employ machine learning and artificial intelligence techniques to categorize them. This research introduces a novel method for the analysis and classification of WSN datasets. The primary objective is to develop an anomaly identification approach that enhances sensor network security and operational efficiency with a good degree of accuracy. To achieve this goal, artificial intelligence method based-on stochastic models are used to create a detection system that learns from existing routing data to identify potential malicious network entries. The proposed approach relies on the principles of the Hidden Markov Model (HMM) and the Gaussian Mixture Model (GMM), a part of artificial intelligence stochastic functions, which incorporate predictive assumptions. In addition, dimensionality reduction is used to select the most pertinent routing features for the training of the system. To assess the effectiveness of our proposed approach, we performed experiments using a custom dataset that represents various network scenarios, including both normal and attacked states. The results demonstrate the performance of the model, achieving a classification score of 92. 18% when using a combination of two HMM and three GMM in the classifier. The proposed method attains a 98% precision value and 95% accuracy, better than performances of SVM, NB, DT and RF methods. This highlights the efficacy of our proposed approach compared to existing research. [ABSTRACT FROM AUTHOR]

The unsupervised anomaly detection in high-dimensional and complex settings poses a formidable challenge. To tackle the challenges associated with the recognition of high-dimensional data, this paper proposes a feedback channel between the Variational Auto Encoder and the Gaussian process to enhance its data feature extraction capabilities. In order to alleviate the impact of common mislabeling issues within unsupervised contexts, the proposed model introduces a method for the selection of representative data. Subsequently, these representative data points are employed as anchors for the construction of a full-data Gaussian process model covering the entire dataset. To further disentangle the intertwined normal and abnormal data, the proposed model employs a Bayesian framework to guide the displacement of each data point within the latent space toward its position with the highest confidence. This strategic approach effectively discriminates between normal and abnormal data. This innovative method significantly enhances data consolidation within the latent space, thus preserving robustness, particularly when processing high-complexity data. Also, the numerical experiments not only validate the method's effectiveness but also showcase the model's robustness. These experiments highlight that even a small number of incorrect labels do not significantly impact the accuracy of anomaly detection. [ABSTRACT FROM AUTHOR]

Motivation The crisis of antibiotic resistance, which causes antibiotics used to treat bacterial infections to become less effective, has emerged as one of the foremost challenges to public health. Identifying the properties of antibiotic resistance genes (ARGs) is an essential way to mitigate this issue. Although numerous methods have been proposed for this task, most of these approaches concentrate solely on predicting antibiotic class, disregarding other important properties of ARGs. In addition, existing methods for simultaneously predicting multiple properties of ARGs fail to account for the causal relationships among these properties, limiting the predictive performance. Results In this study, we propose a causality-guided framework for annotating properties of ARGs, in which causal inference is utilized for representation learning. More specifically, the hidden biological patterns determining the properties of ARGs are described by a Gaussian Mixture Model, and procedure of causal representation learning is used to derive the hidden features. In addition, a causal graph among different properties is constructed to capture the causal relationships among properties of ARGs, which is integrated into the task of annotating properties of ARGs. The experimental results on a real-world dataset demonstrate the effectiveness of the proposed framework on the task of annotating properties of ARGs. Availability and implementation The data and source codes are available in GitHub at https://github.com/David-WZhao/CausalARG. [ABSTRACT FROM AUTHOR]

Optimization of the routing represents an important challenge when considering the rapid development of Wireless Sensor Networks (WSN), which involve efficient energy methods. Applying the effectiveness of a Deep Neural Network (DNN) and a Gaussian Mixture Model (GMM), the present article proposes an innovative method for attaining Energy-Efficient Routing (EER) in WSN. When it comes to dealing with dynamic network issues, conventional routing protocols generally conflict, resulting in unsustainable Energy consumption (EC). By applying algorithms based on data mining to adapt routing selections in an effective procedure, the GMM + DNN methodology that has been developed is able to successfully address this problem. The GMM is a fundamental Feature Extraction (FE) method for accurately representing the features of statistical analysis associated with network parameters like signal frequency, the amount of traffic, and channel states. By learning from previous data collection, the DNN, which relies on these FE, provides improved routing selections, resulting in more efficient use of energy. Since routing paths are constantly optimized to ensure dynamic adaptation, where less energy is used, networks last longer and perform more efficiently. Network simulations highlight the GMM + DNN method's effectiveness and depict how it outperforms conventional routing methods while preserving network connectivity and data throughput. The GMM + DNN's adaptability to multiple network topologies and traffic patterns and its durability make it an efficient EER technique in the diverse WSN context. The GMM + DNN achieves an EC of 0.561 J, outperforming the existing state-of-the-art techniques. [ABSTRACT FROM AUTHOR]

Steel Plate Cold-Rolled Commercial (SPCC) steel is known to have long-term durability. However, it still undergoes corrosion when exposed to corrosive environments. This paper proposes an evaluation method for assessing the corrosion level of SPCC steel samples using eddy current testing (ECT), along with two different machine learning approaches. The objective is to classify the corrosion of the samples into two states: a less corroded state (state-1) and a highly corroded state (state-2). Generative and discriminative models were implemented for classification. The generative classifier was based on the Gaussian mixture model (GMM), while the discriminative model was based on the logistic regression model. The features used in the classification models are the peaks of the perturbated magnetic fields at two different frequencies. The performance of the classifiers was evaluated using metrics such as absolute error, accuracy, precision, recall, and F1 score. The results indicate that the GMM model is more conducive to categorizing states with higher levels of corrosion, while the logistic regression model is helpful in estimating states with lower levels of corrosion. Meanwhile, high classification accuracy can be achieved based on both methods using eddy current testing. [ABSTRACT FROM AUTHOR]

This paper addresses the critical challenge of detecting, separating, and classifying partial discharges in substations. It proposes two solutions: the first involves developing a signal conditioning system to reduce the sampling requirements for PD detection and increase the signal-to-noise ratio. The second approach uses machine learning techniques to separate and classify PD based on features extracted from the conditioned signal. Three clustering algorithms (K-means, Gaussian Mixture Model (GMM), and Mean-shift) and the Support Vector Machine (SVM) method were used for signal separation and classification. The proposed system effectively reduced high-frequency components up to 50 MHz, improved the signal-to-noise ratio, and effectively separated different sources of partial discharges without losing relevant information. An accuracy of up to 93% was achieved in classifying the partial discharge sources. The successful implementation of the signal conditioning system and the machine learning-based signal separation approach opens avenues for more economical, scalable, and reliable PD monitoring systems. [ABSTRACT FROM AUTHOR]

In the realm of road safety and the evolution toward automated driving, Advanced Driver Assistance and Automated Driving (ADAS/AD) systems play a pivotal role. As the complexity of these systems grows, comprehensive testing becomes imperative, with virtual test environments becoming crucial, especially for handling diverse and challenging scenarios. Radar sensors are integral to ADAS/AD units and are known for their robust performance even in adverse conditions. However, accurately modeling the radar's perception, particularly the radar cross-section (RCS), proves challenging. This paper adopts a data-driven approach, using Gaussian mixture models (GMMs) to model the radar's perception for various vehicles and aspect angles. A Bayesian variational approach automatically infers model complexity. The model is expanded into a comprehensive radar sensor model based on object lists, incorporating occlusion effects and RCS-based detectability decisions. The model's effectiveness is demonstrated through accurate reproduction of the RCS behavior and scatter point distribution. The full capabilities of the sensor model are demonstrated in different scenarios. The flexible and modular framework has proven apt for modeling specific aspects and allows for an easy model extension. Simultaneously, alongside model extension, more extensive validation is proposed to refine accuracy and broaden the model's applicability. [ABSTRACT FROM AUTHOR]

The purpose of this paper is twofold. The first part is to introduce relative- $\chi_{\alpha}^{2}$ , Jensen- $\chi_{\alpha}^{2}$ and (p , w)-Jensen- $\chi_{\alpha}^2$ divergence measures and then examine their properties. In addition, we also explore possible connections between these divergence measures and Jensen–Shannon entropy measure. In the second part, we introduce $(p,\eta)$ -mixture model and then show it to be an optimal solution to three different optimization problems based on $\chi_{\alpha}^{2}$ divergence measure. We further study the relative- $\chi_{\alpha}^{2}$ divergence measure for escort and arithmetic mixture densities. We also provide some results associated with relative- $\chi_{\alpha}^{2}$ divergence measure of mixed reliability systems. Finally, to demonstrate the usefulness of the Jensen- $\chi_{\alpha}^{2}$ divergence measure, we apply it to a real example in image processing and present some numerical results. Our findings in this regard show that the Jensen- $\chi_{\alpha}^{2}$ is an effective criteria for quantifying the similarity between two images. [ABSTRACT FROM AUTHOR]

Underwater signal processing is primarily based on sound waves because of the unique properties of water. However, the slow speed and limited bandwidth of sound introduce numerous challenges, including pronounced time-varying characteristics and significant multipath effects. This paper explores a channel estimation method utilizing superimposed training sequences. Compared with conventional schemes, this method offers higher spectral efficiency and better adaptability to time-varying channels owing to its temporal traversal. To ensure success in this scheme, it is crucial to obtain time-varying channel estimation and data detection at low SNRs given that superimposed training sequences consume power resources. To achieve this goal, we initially employ coarse channel estimation utilizing superimposed training sequences. Subsequently, we employ approximate message passing algorithms based on the estimated channels for data detection, followed by iterative channel estimation and equalization based on estimated symbols. We devise an approximate message passing channel estimation method grounded on a Gaussian mixture model and refine its hyperparameters through the expectation maximization algorithm. Then, we refine the channel information based on time correlation by employing an autoregressive hidden Markov model. Lastly, we perform numerical simulations of communication systems by utilizing a time-varying channel toolbox to simulate time-varying channels, and we validate the feasibility of the proposed communication system using experimental field data. [ABSTRACT FROM AUTHOR]

This study investigates the rare-event system reliability analysis with numerous failure regions. A novel method, based on the parallel tempering (PT) and importance sampling (IS) technique, is proposed. Surrogate models (kriging model) are built for the true performance function and a probabilistic classification function is derived to predict the failure regions. The PT algorithm is used to obtain points populating all of the predicted failure regions. Representative points are identified using the k-weighted-means clustering method. A Gaussian mixture model is formulated by the representative points and importance samples are simulated accordingly. The optimal training points are selected to update all surrogate models. In the framework of IS, the terminating criterion for assessing the estimation error of the system failure probability is devised. The learning process is terminated at the appropriate stage. The method is termed active learning kriging–parallel tempering–importance sampling (ALK-PT-IS). Four numerical examples illustrate the effectiveness and precision of this method. [ABSTRACT FROM AUTHOR]

Latent Dirichlet allocation (LDA) is one of the major models used for topic modelling. A number of models have been proposed extending the basic LDA model. There has also been interesting research to replace the Dirichlet prior of LDA with other pliable distributions like generalized Dirichlet, Beta‐Liouville and so forth. Owing to the proven efficiency of using generalized Dirichlet (GD) and Beta‐Liouville (BL) priors in topic models, we use these versions of topic models in our paper. Furthermore, to enhance the support of respective topics, we integrate mixture components which gives rise to generalized Dirichlet mixture allocation and Beta‐Liouville mixture allocation models respectively. In order to improve the modelling capabilities, we use variational inference method for estimating the parameters. Additionally, we also introduce an online variational approach to cater to specific applications involving streaming data. We evaluate our models based on its performance on applications related to text classification, image categorization and genome sequence classification using a supervised approach where the labels are used as an observed variable within the model. [ABSTRACT FROM AUTHOR]

Third‐party automatic weather stations (TPAWS) provide a compelling data source for scientists and practitioners to observe and estimate more accurate fine‐scale atmospheric conditions, including daily maximum and minimum temperature (denoted as Tmax and Tmin, respectively), than the current primary weather observation network can offer. Several uncertainties and errors arise in data from TPAWS as the quality control applied to these stations may be inadequate or ad hoc. In this study, we develop a statistical approach to evaluate the quality of daily Tmax and Tmin observations collected from TPAWS in Australia. Our approach compares a target observation with multiple types of reliable reference data, including neighbouring primary weather observations from the official Bureau of Meteorology of Australia stations, Australian Gridded Climate Data, and numerical weather prediction data. Guided by the operational requirements in terms of automation, interpretability, and simplicity as well as expandability, a separate test is formed for each type of reference data and then all the individual tests are merged to generate a single result based on a Gaussian mixture model that is used to provide the final overall assessment for each TPAWS observation. The overall assessment is made in the form of a p‐value‐based confidence score that measures the difference between the target observation and trusted reference data. Our method is validated by synthetic datasets based on high‐quality observations and is also applied to daily Tmax and Tmin observations from 184 TPAWS owned by the Department of Primary Industries and Regional Development of Western Australia. The framework can be readily applied to different regions with different reliable or trusted data sources. [ABSTRACT FROM AUTHOR]

This paper proposes a dual-channel speech separation method based on spatial estimation via sparse Bayesian inference (SBI) and nonnegative matrix factorization (NMF). The spatial information estimated by traditional compressed sensing (CS) models is insufficient when two microphones receive limited columns of mixed signals. Considering the sparsity of peak values in the cross-correlation spectrum between two received signals, the proposed method builds a new CS model based on cross-correlation spectrum and applies SBI algorithm to solve this model to improve the estimation accuracy of spatial information. Combined the spatial information with the spectral features decomposed by NMF, NMF coefficient matrix masks belonging to individual source are generated for pre-separation. To mitigate retained potential interference components, a post-separation processing stage is designed using an expectation maximization (EM) algorithm based on a Gaussian mixture model (GMM). The estimated spatial information and binary time–frequency masks are used for parameter initialization of the EM algorithm. The experimental results using real-world speech data show that the proposed method can achieve better separation performance compared to various existing methods. [ABSTRACT FROM AUTHOR]

The strawberry appearance is an essential standard for judging the quality, so it is crucial to accurately identify the strawberry appearance quality for intelligent picking. This study proposed a new strawberry appearance quality detection based on unsupervised deep learning. Firstly, using deep learning (Resnet18, Resnet50, and Resnet101) to extract the strawberry image feature information. And using the t-SNE (t-distribution stochastic neighbor embedding) to reduce the feature vectors' dimension. Finally, the unsupervised learning method (Gaussian Mixture Model) was used to cluster strawberries' feature points. The results showed that: (1) the clustering performance based on Resnet101 was effective in 2-dimensional space, the cluster accuracy was 94.89%, and the validation accuracy was 91.79%. (2) The clustering method based on Resnet50 had good performance in the 3-dimensional space, the cluster accuracy was 96.10%, and the validation accuracy was 93.08%. (3) The accuracy of deep features plus RF (random forest) was 95.00% under limited data. Thus this method will promote intelligent picking strawberry equipment and it will overcome the supervised learning drawback that divides image datasets according to prior knowledge. [ABSTRACT FROM AUTHOR]

With the popularity and advancement of 3D point cloud data acquisition technologies and sensors, research into 3D point clouds has made considerable strides based on deep learning. The semantic segmentation of point clouds, a crucial step in comprehending 3D scenes, has drawn much attention. The accuracy and effectiveness of fully supervised semantic segmentation tasks have greatly improved with the increase in the number of accessible datasets. However, these achievements rely on time‐consuming and expensive full labelling. In solve of these existential issues, research on weakly supervised learning has recently exploded. These methods train neural networks to tackle 3D semantic segmentation tasks with fewer point labels. In addition to providing a thorough overview of the history and current state of the art in weakly supervised semantic segmentation of 3D point clouds, a detailed description of the most widely used data acquisition sensors, a list of publicly accessible benchmark datasets, and a look ahead to potential future development directions is provided. [ABSTRACT FROM AUTHOR]

Although robots have been widely used in a variety of fields, the idea of enabling them to perform multiple tasks in the same way that humans do remains a difficulty. To solve this, we investigate the learning from demonstration (LFD) system with our independently designed symmetrical humanoid dual-arm robot. We present a novel action feature matching algorithm. This algorithm accurately transforms human demonstration data into task models that robots can directly execute, considerably improving LFD's generalization capabilities. In our studies, we used motion capture cameras to capture human demonstration actions, which included combinations of simple actions (the action layer) and a succession of complicated operational tasks (the task layer). For the action layer data, we employed Gaussian mixture models (GMM) for processing and constructing an action primitive library. As for the task layer data, we created a "keyframe" segmentation method to transform this data into a series of action primitives and build another action primitive library. Guided by our algorithm, the robot successfully imitated complex human tasks. Results show its excellent task learning and execution, providing an effective solution for robots to learn from human demonstrations and significantly advancing robot technology. [ABSTRACT FROM AUTHOR]

Clustering algorithms are usually iterative procedures. In particular, when the clustering algorithm aims to optimise an objective function like in k-means clustering or Gaussian mixture models, iterative heuristics are required due to the high non-linearity of the objective function. This implies higher computational costs and the risk of finding only a local optimum and not the global optimum of the objective function. In this paper, we demonstrate that in the case of one-dimensional clustering with one main and one noise cluster, one can formulate an objective function, which permits a closed-form solution with no need for an iteration scheme and the guarantee of finding the global optimum. We demonstrate how such an algorithm can be applied in the context of laboratory medicine as a method to estimate reference intervals that represent the range of "normal" values. [ABSTRACT FROM AUTHOR]

Finite Gaussian mixture models are powerful tools for modeling distributions of random phenomena and are widely used for clustering tasks. However, their interpretability and efficiency are often degraded by the impact of redundancy and noise, especially on high-dimensional datasets. In this work, we propose a generative graphical model for parsimonious modeling of the Gaussian mixtures and robust unsupervised learning. The model assumes that the data are generated independently and identically from a finite mixture of robust factor analyzers, where the features' salience is adjusted by an active set of latent factors to allow a violation of the local independence assumption. For the model inference, we propose a structured variational Bayes inference framework to realize simultaneous clustering, model selection and outlier processing. Performance of the proposed algorithm is evaluated by conducting experiments on artificial and real-world datasets. Moreover, an application on the high-dimensional machine learning task of handwritten alphabet recognition is introduced. [ABSTRACT FROM AUTHOR]

The study of process mineralogy plays a very important role in the field of mineral processing and metallurgy, in which the measurement of mineral-embedded particle size is one of the main research areas. The manual measurement method using a microscope has many problems, such as heavy workload and low measurement accuracy. In order to solve this problem, this paper proposes a Gaussian mixture model based on an expectation maximization (EM) algorithm to measure the embedded particle sizes of minerals of polished metal sections. Experiments are here performed on the polished section images of ilmenite and pyrite, and we compared the results with a microscope. The experimental results show that the proposed method has higher precision and accuracy in measuring the embedded particle sizes of metal minerals. [ABSTRACT FROM AUTHOR]

Practical engineering problems often involve stochastic uncertainty, which can cause substantial variations in the response of engineering products or even lead to failure. The coupling and propagation of uncertainty play a crucial role in this process. Hence, it is imperative to quantify, propagate and control stochastic uncertainty. Different from most traditional uncertainty propagation methods, the proposed method employs Gaussian splitting method to divide the input random variables into Gaussian mixture models. These GMMs have a limited number of components with very small variances. As a result, the input Gaussian components can be conveniently propagated to the response and remain Gaussian distributions after nonlinear uncertainty propagation, which is able to provide an effective method for high-precision nonlinear uncertainty propagation. Firstly, the probability density function of input random variable is reconstructed by Gaussian mixture models. Secondly, the K-value criterion is proposed for selecting split direction, taking into account both the nonlinearity and variance. The components of input random variables are then divided into a Gaussian mixture model with small variance along the direction determined by the K-value. Thirdly, the individual components of the Gaussian mixture model are propagated one by one to obtain the probability density function of the response. Finally, the convergence criterion based on Shannon entropy is developed to ensure the accuracy of uncertainty propagation. The efficacy of the method is verified using three numerical examples and two engineering examples. [ABSTRACT FROM AUTHOR]

Automatic crack segmentation plays an essential role in maintaining the structural health of buildings and infrastructure. Despite the success in fully supervised crack segmentation, the costly pixel-level annotation restricts its application, leading to increased exploration in weakly supervised crack segmentation (WSCS). However, WSCS methods inevitably bring in noisy pseudo-labels, which results in large fluctuations. To address this problem, we propose a novel confidence-aware co-training (CAC) framework for WSCS. This framework aims to iteratively refine pseudo-labels, facilitating the learning of a more robust segmentation model. Specifically, a co-training mechanism is designed and constructs two collaborative networks to learn uncertain crack pixels, from easy to hard. Moreover, the dynamic division strategy is designed to divide the pseudo-labels based on the crack confidence score. Among them, the high-confidence pseudo-labels are utilized to optimize the initialization parameters for the collaborative network, while low-confidence pseudo-labels enrich the diversity of crack samples. Extensive experiments conducted on the Crack500, DeepCrack, and CFD datasets demonstrate that the proposed CAC significantly outperforms other WSCS methods. [ABSTRACT FROM AUTHOR]

One of the common nervous system diseases in older adults is Alzheimer's and epilepsy, and the possibility of occurrence increases with age. The chances of seizure are high for patients with mild cognitive impairment and Alzheimer's disease. So, there is a bidirectional association between Alzheimer's and epilepsy, as both affect the neurodegenerative processes. Electroencephalogram (EEG) is a possible non-invasive measurement technique widely used to measure the variations in brain signals. EEG signal is analyzed to discriminate the Alzheimer and epilepsy. Numerous research works evaluated the clinical relevance of Alzheimer's and epilepsy. Specifically, machine learning-based evaluation models developed recently bring the facts by extracting features from the EEG signals. However, machine learning-based models lag in performance due to high dimensional EEG features. For initial feature selection particle swarm optimization is included in the proposed model and to reduce the computation complexity of the classifier, kernel PCA is incorporated for dimensionality reduction. Experimentations using benchmark Bon and Dementia datasets confirms the proposed model better performances in terms of precision, recall, f1-score and accuracy. The attained accuracy of 94% is much better than existing Gaussian Mixture Model (GMM), Relevance Vector Machine (RVM), Support Vector Machine (SVM), and Artificial Neural Network (ANN) methods. [ABSTRACT FROM AUTHOR]

In this paper, a tight lower bound for the differential entropy of the Gaussian mixture model is presented. First, the probability model of mixed Gaussian distribution that is created by mixing both discrete and continuous random variables is investigated in order to represent symmetric bimodal Gaussian distribution using the hyperbolic cosine function, on which a tighter upper bound is set. Then, this tight upper bound is used to derive a tight lower bound for the differential entropy of the Gaussian mixture model introduced. The proposed lower bound allows to maintain its tightness over the entire range of the model’s parameters and shows more tightness when compared with other bounds that lose their tightness over certain parameter ranges. The presented results are then extended to introduce a more general tight lower bound for asymmetric bimodal Gaussian distribution, in which the two modes have a symmetric mean but differ in terms of their weights. [ABSTRACT FROM AUTHOR]

Visual quality measures (VQMs) are designed to support analysts by automatically detecting and quantifying patterns in visualizations. We propose a new VQM for visual grouping patterns in scatterplots, called ClustML, which is trained on previously collected human subject judgments. Our model encodes scatterplots in the parametric space of a Gaussian Mixture Model and uses a classifier trained on human judgment data to estimate the perceptual complexity of grouping patterns. The numbers of initial mixture components and final combined groups quantify visual cluster patterns in scatterplots. It improves on existing VQMs, first, by better estimating human judgments on two-Gaussian cluster patterns and, second, by giving higher accuracy when ranking general cluster patterns in scatterplots. We use it to analyze kinship data for genome-wide association studies, in which experts rely on the visual analysis of large sets of scatterplots. We make the benchmark datasets and the new VQM available for practical use and further improvements. [ABSTRACT FROM AUTHOR]

Habitat structural complexity (HSC) is a key variable to help interpret ecological patterns and processes among different ecosystems. Existing metrics used to quantify HSC often, however, result in insufficient or biased representation of structural complexity. As such, our understanding of how HSC affects biodiversity and related ecological patterns is often limited by these measures.Recent advances in photogrammetry and computer vision have enabled the ability to reconstruct 3D habitats with high efficiency and accuracy. Point clouds, for example, better represent the structural properties of target objects with reduced sampling bias as compared to traditional formats like 2.5D raster (i.e. DEM, DSM, DTM). The analysis of point clouds is, however, limited by the lack of readily available packages with mathematically well‐defined metrics for ecologists to help interpret relevant HSC properties.To address this gap, novel metrics are provided in the present Python package HSC3D version 0.2.0, which allows quantification of structural complexity of targeted habitats based on photogrammetry point clouds. This package is designed to help ecologists better describe the structural characteristics of a specific habitat and is bundled with visualisation functions to help interpret the computational processes used.To demonstrate functions implemented in HSC3D we use a case study that compares the structural complexity differences formed by two intertidal ecosystem engineers, mussels and oysters. Results indicate that HSC3D is a versatile and easily adapted package, which can provide both ecological and analytical insights on photogrammetry point clouds which should prove a useful tool for ecologists wanting to quantify HSC. [ABSTRACT FROM AUTHOR]

Accurate positioning is one of the essential requirements for numerous applications of remote sensing data, especially in the event of a noisy or unreliable satellite signal. Toward this end, we present a novel framework for aircraft geo-localization in a large range that only requires a downward-facing monocular camera, an altimeter, a compass, and an open-source Vector Map (VMAP). The algorithm combines the matching and particle filter methods. Shape vector and correlation between two building contour vectors are defined, and a coarse-to-fine building vector matching (CFBVM) method is proposed in the matching stage, for which the original matching results are described by the Gaussian mixture model (GMM). Subsequently, an improved resampling strategy is designed to reduce computing expenses with a huge number of initial particles, and a credibility indicator is designed to avoid location mistakes in the particle filter stage. An experimental evaluation of the approach based on flight data is provided. On a flight at a height of 0.2 km over a flight distance of 2 km, the aircraft is geo-localized in a reference map of 11,025 km2 using 0.09 km² aerial images without any prior information. The absolute localization error is less than 10 m. [ABSTRACT FROM AUTHOR]

Clustering holds significant utility across a spectrum of several domains, including pattern recognition, image analysis, customer analytics, market segmentation, social network analysis, and numerous other areas. The main advantages of image clustering are its degree of freedom regarding data labeling and the lack of training and model deployment, which makes them suitable for the overall study's purpose of land cover segmentation and deforestation monitoring. In previous work, the Gaussian Mixture Model (GMM) technique has been established as the best option. Due to the necessity of implementing the algorithm on light unmanned airborne platforms for fast deforestation monitoring, the high resources and long computation time became an issue. This paper proposes several cost-efficient GMM clustering algorithms based on discrete transforms traditionally used for image compression. The results will show that the proposed methods maintain the clustering output quality while drastically decreasing the computation time and also lowering the memory needed to perform. [ABSTRACT FROM AUTHOR]

Connected and Autonomous Vehicles (CAVs) technology has the potential to transform the transportation system. Although these new technologies have many advantages, the implementation raises significant concerns regarding safety, security, and privacy. Anomalies in sensor data caused by errors or cyberattacks can cause severe accidents. To address the issue, this study proposed an innovative anomaly detection algorithm, namely the LSTM Autoencoder with Gaussian Mixture Model (LAGMM). This model supports anomalous CAV trajectory detection in the real-time leveraging communication capabilities of CAV sensors. The LSTM Autoencoder is applied to generate low-rank representations and reconstruct errors for each input data point, while the Gaussian Mixture Model (GMM) is employed for its strength in density estimation. The proposed model was jointly optimized for the LSTM Autoencoder and GMM simultaneously. The study utilizes realistic CAV data from a platooning experiment conducted for Cooperative Automated Research Mobility Applications (CARMAs). The experiment findings indicate that the proposed LAGMM approach enhances detection accuracy by 3% and precision by 6.4% compared to the existing state-of-the-art methods, suggesting a significant improvement in the field. [ABSTRACT FROM AUTHOR]