Search Results (7,249)

Neural Architecture Search (NAS) has found applications in various areas of computer vision, including image recognition and object detection. An increasing number of algorithms, such as ENAS (Efficient Neural Architecture Search via Parameter Sharing) and DARTS (Differentiable Architecture Search), have been applied to NAS. Nevertheless, the current Training-free NAS methods continue to exhibit unreliability and inefficiency. This paper introduces a training-free prune-based algorithm called TTNAS (True-Skill Training-Free Neural Architecture Search), which utilizes a Bayesian method (true-skill algorithm) to combine multiple indicators for evaluating neural networks across different datasets. The algorithm demonstrates highly competitive accuracy and efficiency compared to state-of-the-art approaches on various datasets. Specifically, it achieves 93.90% accuracy on CIFAR-10, 71.91% accuracy on CIFAR-100, and 44.96% accuracy on ImageNet 16-120, using 1466 GPU seconds in NAS-Bench-201. Additionally, the algorithm exhibits improved adaptation to other datasets and tasks. Full article

Epidemiological approaches for examining human health responses to environmental exposures in observational studies frequently address confounding by employing advanced matching techniques and statistical methods grounded in conditional likelihood. This study incorporates a recently developed Bayesian hierarchical spatiotemporal model within a conditional logistic regression framework to capture the heterogeneous effects of environmental exposures in a case-crossover (CCO) design. Spatial and temporal dependencies are modeled through random effects incorporating multivariate conditional autoregressive priors. Flexible frailty structures are introduced to explore strategies for managing temporal variables. Parameter estimation and inference are conducted using a Monte Carlo Markov chain method within a Bayesian framework. Model fit and optimal model selection are evaluated using the deviance information criterion. Simulations assess and compare model performance across various scenarios. Finally, the approach is illustrated with workers’ compensation claims data from New York and Florida to examine spatiotemporal heterogeneity in hospitalization rates related to heat prostration. Full article

Frogeye leaf spot on soybeans is traditionally considered as a southern disease in the United States but its impact in North Central USA has been rising in recent years. In this study, we investigated the population structure and mating type distribution in the C. sojina population from Indiana, USA. Based on 27 single nucleotide polymorphism markers, 49 multi-locus genotypes (MLGs) were identified in 234 isolates collected from 29 counties in Indiana in 2020. Bayesian analysis grouped the 49 MLGs into three clusters. This grouping was supported by principal coordinate analysis and, in large part, by the unweighted pair group method with arithmetic mean and minimal spanning tree. Only one mating-type idiomorph was found in each isolate and in each MLG. The MAT1-1 idiomorph was found in 22 MLGs and the MAT1-2 idiomorph was found in 27 MLGs. Based on clone-corrected data, the distribution of mating-type idiomorphs did not deviate significantly from 1:1 ratio in Indiana as a whole and in 22 out of 24 counties where two or more MLGs were found. Thirty MLGs contained QoI-resistant isolates and 22 MLGs contained QoI-sensitive isolates, with three MLGs containing both types of isolates. MLG1, the most common MLG with 90 isolates, contained mostly QoI-resistant isolates. Interestingly, MLG1 was also the dominant genotype in the Tennessee population collected in 2015, suggesting that MLG1 has been a dominant genotype in a wider region for many years. Based on the standard index of association ($\overline{r}$_d), the Indiana population as a whole was in significant linkage disequilibrium. However, in five out of 16 counties where three or more MLGs were found, the null hypothesis of linkage equilibrium was not rejected. Tests of linkage disequilibrium between locus pairs showed that 33.3% of locus pairs on the same contigs were in significant disequilibrium and 17.7% of locus pairs on different contigs were in significant disequilibrium. The possibility of a cryptic sexual stage was discussed. Full article

Background: Pandemic outbreaks have emerged as a significant global threat, with the potential to cause waves of infections that challenge public health systems and disrupt societal norms. Understanding the underlying behavior of disease transmission can be of great use in the design of informed and timely public health policies. It is very common for many contagious diseases not to have actual incidence but rather incidence in a given subgroup. For example, in Spain, as of 28 March 2022, the incidence of COVID-19 in people under 60 years of age is not registered. Methods: This work provides a Bayesian methodology to model the incidence of any infectious disease in the general population when its cases are only registered in a specific subgroup of that population. The case study used was the coronavirus disease (COVID-19), with data for 52 Spanish provinces during the period of 1 January 2020 to 29 August 2022. Results: Explicitly, two multilevel models were proposed, one for people over or of 60 years of age and the other for people under 60 years of age. Performance of the models was 5.9% and 12.7% MAPE, respectively. Conclusions: Despite the limitations of the data and the complexity and uncertainty in the propagation of COVID-19, the models were able to fit the data well and predict incidence with very low MAPE. Full article

Many machine learning (ML)-based detection methods for interpreting ground-penetrating radar (GPR) data of concrete tunnels require extensive labeled damage-state data for model training, limiting their practical use in concealed damage detection of in-service tunnels. This study presents a probabilistic, data-driven method for GPR-based damage detection, which exempts the requirement in the training process of supervised ML models. The approach involves extracting a radar feature vector (RFV), building a Bayesian baseline model with healthy data, and quantifying damage severity with the Bayes factor. The RFV is a complex vector obtained by radargram data fusion. Bayesian regression is applied to build a model for the relationship between real and imaginary parts of the RFV. The Bayes factor is employed for defect identification and severity assessment, by quantifying the difference between the RFV built with new observations and the baseline RFV predicted by the baseline model with new input. The probability of damage is calculated to reflect the influence of uncertainties on the detection result. The effectiveness of the proposed method is validated through simulated data with random noise and physical model tests. This method facilitates GPR-based hidden damage detection of in-service tunnels when lacking labeled damage-state data in the model training process. Full article

In this research, we present a new distribution, which is the bivariate alpha power Burr-XII distribution, based on the alpha power Burr-XII distribution. We thoroughly examine the key features of our newly developed bivariate model. We introduce a new class of bivariate models, which are built with the copula function. The statistical properties of the proposed distribution, such as conditional distributions, conditional expectations, marginal distributions, moment-generating functions, and product moments were studied. This was accomplished with two datasets of real data that came from two distinct devices. We employed Bayesian, maximum likelihood estimation, and least squares estimation strategies to obtain estimated points and intervals. Additionally, we generated bootstrap confidence intervals and conducted numerical analyses using the Markov chain Monte Carlo method. Lastly, we compared this novel bivariate distribution’s performance to earlier bivariate models, to determine how well it fit the real data. Full article

Serious crime modelling typically needs to be undertaken securely behind a firewall where police knowledge and capabilities remain undisclosed. Data informing an ongoing incident are often sparse; a large proportion of relevant data only come to light after the incident culminates or after police intervene—by which point it is too late to make use of the data to aid real-time decision-making for the incident in question. Much of the data that are available to the police to support real-time decision-making are highly confidential and cannot be shared with academics, and are therefore missing to them. In this paper, we describe the development of a formal protocol where a graphical model is used as a framework for securely translating a base model designed by an academic team to a fully embellished model for use by a police team. We then show, for the first time, how libraries of these models can be built and used for real-time decision support to circumvent the challenges of data missingness seen in such a secure environment through the ability to match ongoing plots to existing models within the library.The parallel development described by this protocol ensures that any sensitive information collected by police and missing to academics remains secured behind a firewall. The protocol nevertheless guides police so that they are able to combine the typically incomplete data streams that are open source with their more sensitive information in a formal and justifiable way. We illustrate the application of this protocol by describing how a new entry—a suspected vehicle attack—can be embedded into such a police library of criminal plots. Full article

Nitrate is among the most widely occurring contaminants in groundwater on a global scale, posing a serious threat to drinking water supplies. With the advancement of urbanization and mountainous agriculture, the nitrate in the groundwater of Wanzhou District in the Three Gorges Reservoir Area has formed a complex combination of pollution sources. To more accurately identify the sources of nitrate in groundwater, this study integrates hydrochemical methods and environmental isotope techniques to analyze the sources and transformation processes in shallow groundwater nitrate under different land-use types. Furthermore, the Bayesian isotope mixing model (MixSAIR) is employed to calculate the contribution rates in various nitrate sources. The results indicate that nitrate is the primary form of inorganic nitrogen in shallow groundwater within the study area, with nitrate concentrations in cultivated groundwater generally higher than those in construction land and forest land. The transformation process of nitrate is predominantly nitrification, with little to no denitrification observed. In cultivated shallow groundwater, nitrate mainly originates from chemical fertilizers (36.3%), sewage and manure (35.4%), and soil organic nitrogen (24.7%); in forested areas, nitrate primarily comes from atmospheric precipitation (35.3%), chemical fertilizers (31.3%), and soil organic nitrogen (22.1%); while in constructed areas, nitrate mainly derives from chemical fertilizers (46.0%) and sewage and manure (32.2%). These results establish a scientific foundation for formulating groundwater pollution control and management strategies in the region and serve as a reference for identifying nitrate sources in groundwater in regions with comparable hydrogeological features and pollution profiles. Full article

Human activities, especially industrial production and urbanization, have significantly affected vegetation cover, water resource cycles, climate change, and biodiversity in the Qinling-Daba Mountain region and its surrounding areas. These activities contribute to complex and lasting impacts on ecological vulnerability. The Qinling Mountain region exhibits a complex interaction with human activities. The current research on the ecological vulnerability of the Qinling Mountain region primarily focuses on spatial distribution and the driving factors. This study innovatively applies the VSD assessment and Bayesian networks to systematically evaluate and simulate the ecological vulnerability of the study area over the past 20 years, which indicates that the integration of the VSD model with the Bayesian network model enables the simulation of dynamic relationships and interactions among various factors within the study areas, providing a more accurate assessment and prediction of ecosystem responses to diverse changes from a dynamic perspective. The key findings are as follows. (1) Areas of potential and slight vulnerability are concentrated in the Qinling-Daba mountainous regions. Over the past 20 years, areas of extreme and high vulnerability have significantly decreased, while areas of potential vulnerability and slight vulnerability have increased. (2) The key factors impacting ecological vulnerability during this period included industrial water use, SO₂ emissions, industrial wastewater, and ecological water use. (3) Areas primarily hindering the transition to potential vulnerability are concentrated in well-developed small urban regions within basins. Furthermore, natural factors like altitude and temperature, which cannot be artificially regulated, are the major impediments to future ecological restoration. Therefore, this paper recommends natural restoration strategies based on environmental protection and governance strategies that prioritize green development as complementary measures. The discoveries of the paper provide a novel analytical method for the study of ecological vulnerability in mountainous areas, offering valuable insights for enhancing the accuracy of ecological risk prediction, fostering the integration of interdisciplinary research, and optimizing environmental governance and protection strategies. Full article

Individual physiotherapy is crucial in treating patients with various pain and health issues, and significantly impacts abdominal surgical outcomes and further medical problems. Recent technological and artificial intelligent advancements have equipped healthcare professionals with innovative tools, such as sensor systems and telemedicine equipment, offering groundbreaking opportunities to monitor and analyze patients’ physical activity. This paper investigates the potential applications of mobile accelerometers in evaluating the symmetry of specific rehabilitation exercises using a dataset of 1280 tests on 16 individuals in the age range between 8 and 75 years. A comprehensive computational methodology is introduced, incorporating traditional digital signal processing, feature extraction in both time and transform domains, and advanced classification techniques. The study employs a range of machine learning methods, including support vector machines, Bayesian analysis, and neural networks, to evaluate the balance of various physical activities. The proposed approach achieved a high classification accuracy of 90.6% in distinguishing between left- and right-side motion patterns by employing features from both the time and frequency domains using a two-layer neural network. These findings demonstrate promising applications of precise monitoring of rehabilitation exercises to increase the probability of successful surgical recovery, highlighting the potential to significantly enhance patient care and treatment outcomes. Full article

The brain is a biological system comprising nerve cells and orchestrates its embodied agent’s perception, behavior, and learning in dynamic environments. The free-energy principle (FEP) advocated by Karl Friston explicates the local, recurrent, and self-supervised cognitive dynamics of the brain’s higher-order functions. In this study, we continue to refine the FEP through a physics-guided formulation; specifically, we apply our theory to synaptic learning by considering it an inference problem under the FEP and derive the governing equations, called Bayesian mechanics. Our study uncovers how the brain infers weight changes and postsynaptic activity, conditioned on the presynaptic input, by deploying generative models of the likelihood and prior belief. Consequently, we exemplify the synaptic efficacy in the brain with a simple model; in particular, we illustrate that the brain organizes an optimal trajectory in neural phase space during synaptic learning in continuous time, which variationally minimizes synaptic surprisal. Full article

This study aims to explore the efficacy of Bayesian Neural Networks (BNNs) in enhancing predictive modeling for healthcare applications. Advancements in artificial intelligence have significantly improved predictive modeling capabilities, with BNNs offering a probabilistic framework that addresses the inherent uncertainty and variability in healthcare data. This study demonstrates the real-world applicability of BNNs through three key case studies: personalized diabetes treatment, early Alzheimer’s disease detection, and predictive modeling for HbA1c levels. By leveraging the Bayesian approach, these models provide not only enhanced predictive accuracy but also uncertainty quantification, a critical factor in clinical decision making. While the findings are promising, future research should focus on optimizing scalability and integration for real-world applications. This work lays a foundation for future studies, including the development of rating scales based on BNN predictions to improve clinical outcomes. Full article

This study focuses on the Siberian roe deer population in the Greater Khingan Mountains, Northeast China. The cold temperate forest ecosystem in this area is distinctive. The Siberian roe deer is a crucial ecological indicator species, and its living conditions hold significant importance for ecological balance. From the winter of 2019 to 2022, 269 fecal samples of Siberian roe deer were collected from four protected areas in the northern part of the Greater Khingan Mountains, Heilongjiang Province. These samples were comprehensively analyzed using mitochondrial DNA and microsatellite markers, combined with conservation genetics evaluation methods. The results revealed that 244 individuals were identified in the fecal samples. The results of a Cyt b genetic analysis of the samples indicated that the haplotype and nucleotide diversity were 88.1% and 20.3%, respectively. The evaluation of 14 pairs of microsatellite loci showed that the average number of alleles was 11.2, and the average expected and observed heterozygosity were 0.672 and 0.506, respectively. Therefore, the overall genetic diversity level is high, but some populations of Siberian roe deer are at risk. AMOVA analysis and STRUCTURE Bayesian clustering confirmed the existence of obvious genetic differentiation among the populations. Historical studies have shown that the HZ and SH populations underwent the earliest diffusion events, and the BJC and SL populations also exhibited related signs (HZ: Huzhong Nature Reserve in the Greater Khingan Mountains; SH: Shuanghe National Nature Reserve in Heilongjiang Province; BJC: Heilongjiang Beijicun National Nature Reserve; SL: Songling District in Heilongjiang Province). Mismatch distribution and neutral tests indicated no expansion events or bottleneck effects in the population, and the inbreeding coefficient was positive, suggesting the possibility of inbreeding. The development potential of the population in the future varies among the various local populations. This study supports the biodiversity of Siberian roe deer at the genetic level to save the population and provides important scientific basis and reference for the protection and management of Siberian roe deer. Full article

Emotion classification is a challenge in affective computing, with applications ranging from human–computer interaction to mental health monitoring. In this study, the classification of emotional states using electroencephalography (EEG) data were investigated. Specifically, the efficacy of the combination of various feature selection methods and hyperparameter tuning of machine learning algorithms for accurate and robust emotion recognition was studied. The following feature selection methods were explored: filter (SelectKBest with analysis of variance (ANOVA) F-test), embedded (least absolute shrinkage and selection operator (LASSO) tuned using Bayesian optimization (BO)), and wrapper (genetic algorithm (GA)) methods. We also executed hyperparameter tuning of machine learning algorithms using BO. The performance of each method was assessed. Two different EEG datasets, EEG Emotion and DEAP Dataset, containing 2548 and 160 features, respectively, were evaluated using random forest (RF), logistic regression, XGBoost, and support vector machine (SVM). For both datasets, the experimented three feature selection methods consistently improved the accuracy of the models. For EEG Emotion dataset, RF with LASSO achieved the best result among all the experimented methods increasing the accuracy from 98.78% to 99.39%. In the DEAP dataset experiment, XGBoost with GA showed the best result, increasing the accuracy by 1.59% and 2.84% for valence and arousal. We also show that these results are superior to those by the previous other methods in the literature. Full article

Group preferences are crucial for Inspirational Solutions of Automotive Design (ISAD). However, sparse individual purchase behavior hinders the identification of group preferences. Therefore, a novel inspiration recommendation (IR) system based on multi-level mining of user behavior data is proposed. Firstly, the K-means algorithm is employed to cluster users based on a variety of features. The fixed association rule is then applied to filter and identify relevant subsets, forming the foundational basis for constructing a user portrait. The Nonlinear Bayesian Personalized Ranking (NBPR) is constructed to explore common preferences using explicit feedback. Finally, the item preference matrix is enriched with implicit feedback to compile a comprehensive recommendation list that caters to group preferences. Using a multi-user joint evaluation approach, we compare the performance of IR with baseline models across multiple metrics. This comparison demonstrates the robust reliability of the IR system and its ability to prioritize ISAD with preference-aligned groups. Our research overcomes data sparsity in the automotive recommendation system, providing a new method for embedding human elements in decision support systems. Full article