Comparison Of Models Research Articles

Notes on the Practical Application of Nested Sampling: MultiNest, (Non)convergence, and Rectification

Nested sampling is a promising tool for Bayesian statistical analysis because it simultaneously performs parameter estimation and facilitates model comparison. MultiNest is one of the most popular nested sampling implementations, and has been applied to a wide variety of problems in the physical sciences. However, MultiNest results, like those of any sampling tool, can be unreliable, and accompanying convergence tests are a component of any analysis. Using analytically tractable test problems, I illustrate how MultiNest, when applied without rigorously chosen hyperparameters, (1) can produce systematically erroneous estimates of the Bayesian evidence, which are more significantly biased for problems of higher dimensionality; (2) can derive posterior estimates with errors on the order of ~100%; (3) can, particularly when sampling noisy likelihood functions, systematically underestimate posterior widths. Furthermore, I show how MultiNest, thanks to the advantageous speed at which it explores parameter space, can also be used to jump-start Markov chain Monte Carlo sampling or more rigorous nested sampling techniques, potentially accelerating more robust measurements of posterior distributions and Bayesian evidences, and overcoming the challenge of Markov chain Monte Carlo initialization.

Research on Convergence Media Ecological Model Based on Blockchain

Currently, the media industry is in the rapid development stage of media integration, which has brought about great changes in content production mode, presentation form, communication mechanism, operation and maintenance management, etc. At the same time, it is also faced with problems such as difficult information traceability, declining industry credibility, low data circulation quality and efficiency, difficult data security and user privacy protection, etc. Utilizing blockchain’s characteristics can solve these problems that the media industry is currently facing. This paper designs a convergence media ecology model based on blockchain (CMEM-BC), focusing on the basic elements of the model, node operation and maintenance system, node management mechanism, value circulation mechanism, and storage mechanism, trying to establish a decentralized, traceable, and immutable convergence media ecosystem. On this basis, this paper summarizes the ecological framework and ecological model of CMEM-BC. Finally, the paper describes the verification of the effectiveness of CMEM-BC in key links through simulation experiments, verifying that CMEM-BC has high originality and is more suitable for the application of convergence media ecology through model analysis and comparison.

Exploring the effect of built environment on spatiotemporal evolution of traffic congestion using a novel GTWR model: a case study of Hefei, China

ABSTRACT Understanding the impact of the built environment on traffic congestion can provide reliable references for alleviating traffic congestion. Previous research has explored traffic congestion evolution but often overlooks two key aspects: the frequency of congestion state updates and the spatial-temporal impact of built environment features. Based on data from Hefei, China, we propose a novel geographically and temporally weighted regression (GTWR) model that integrates temporal variables and spatial grids into the traditional GWR model. The empirical results show that public facilities most significantly impact congestion during morning peak hours, scenic spots during afternoon peaks, and motorcycle services in the evening. The study also reveals the rules for the spatiotemporal impact of the built environment on traffic congestion. Finally, the comparison of models shows that the GTWR model outperforms the OLS and GWR models. The findings can guide traffic managers in creating targeted strategies to enhance transportation system efficiency.

Timed Genetic Process Mining for Robust Tracking of Processes under Incomplete Event Log Conditions

In process mining, an event log is a structured collection of recorded events that describes the execution of processes within an organization. The completeness of event logs is crucial for ensuring accurate and reliable process models. Incomplete event logs, which can result from system errors, manual data entry mistakes, or irregular operational patterns, undermine the integrity of these models. Addressing this issue is essential for constructing accurate models. This research aims to enhance process model performance and robustness by transforming incomplete event logs into complete ones using a process discovery algorithm. Genetic process mining, a type of process discovery algorithm, is chosen for its ability to evaluate multiple candidate solutions concurrently, effectively recovering missing events and improving log completeness. However, the original form of the genetic process mining algorithm is not optimized for handling incomplete logs, which can result in incorrect models being discovered. To address this limitation, this research proposes a modified approach that incorporates timing information to better manage incomplete logs. By leveraging timing data, the algorithm can infer missing events, leading to process tracking and reconstruction which is more accurate. Experimental results validate the effectiveness of the modified algorithm, showing higher fitness and precision scores, improved process model comparisons, and a good level of coverage without errors. Additionally, several advanced metrics for conformance checking are presented to further validate the process models and event logs discovered by both algorithms.

Comparison of Higher-Order Beam Theory and First-Order Beam Theory Models on FGM Beam Element for Static Analysis

HOBT-Higher Order Beam Theory is the developing theory in the Finite Element Method (FEM), which considers the higher-order variation of transverse shear strain; therefore, the shear correction factor is not required. This paper uses beam elements and implements them to higher order beam theory developed by Vo et al., with two independent variables of bending and shear displacement, using third-order polynomial functions. This paper shows the contribution of the shear strain part as one of the independent variables. Static analyses are presented to obtain a comparative result with the First Order Beam Theory (FOBT) application on the imposition of different boundary conditions to ensure reliable results.

Deep learning–based statistical robustness evaluation of intensity-modulated proton therapy for head and neck cancer

A dense, dilated 3D U-net was trained to predict the 5th and 95th percentile distributions of planning dose plan using the nominal dose and planning CT images. The data set comprised proton therapy plans for 582 H&N cancer patients. Ground truth percentile values were estimated for each patient through 600 dose recalculations, representing randomly sampled uncertainty scenarios. The comprehensive comparisons of different models were conducted for H&N cancer patients, considering those with and without a beam mask (BM) and diverse beam configurations, including varying beam angles, couch angles, and beam numbers. The performance of our model trained based on a mixture of patients with H&N and prostate cancer was also assessed in contrast with models trained based on data specific for patients with cancer at either site. The DL-based model's predictions of percentile dose distributions exhibited excellent agreement with the ground truth dose distributions. The average gamma index with 2 mm/2%, consistently exceeded 97% for both 5th and 95th percentile dose volumes. Mean dose-volume histogram error analysis revealed that predictions from the combined training set yielded mean errors and standard deviations that were generally similar to those in the specific patient training data sets. Our proposed DL-based method for evaluation of the robustness of proton therapy plans provides precise, rapid predictions of percentile dose for a given confidence level regardless of the beam arrangement and cancer site. This versatility positions our model as a valuable tool for evaluating the robustness of proton therapy across various cancer sites.

Predicting image quality of forensic footwear impressions

The ability of a footwear examiner to confidently discern features of importance in a forensic examination is directly related to impression quality. As a result, quality directly impacts the strength an examiner can ascribe to any opinion of source attribution. Despite the importance of image quality during both the analysis and comparison phases of an examination, there is limited research on the estimation, variation, and prediction of footwear impression quality. In response, this study aims to develop a methodology for evaluating footwear impression quality by regressing image features against subjective judgments of quality. Using a dataset of more than 450 impressions evaluated by more than 40 participants, estimates of intra- and inter-rater consistency were computed. After identifying reliable raters, matrix completion was performed, thereby permitting data imputation. This was based on an approximate 90:10 split between training and testing data, ultimately resulting in 6,000 quality predictions, which exhibited a mean agreement with ground truth > 95%. Each image in the dataset was then reduced to a 10-dimensional feature vector describing image attributes such as complexity, contrast, sharpness, and noise. These features were used to train multiple ordinal and multinomial regression models aimed at predicting image quality. Model comparison resulted in an overall optimal accuracy of approximately 75% when comparing predicted/modeled quality against the ground truth of subjective rater opinions. The resulting semi-automated, reference-free and numerical prediction tool exhibits reasonable success for impression quality prediction when presented with the types of challenging and domain-specific images often encountered in forensic footwear comparisons. The outcomes and challenges associated with this investigation provide a foundation upon which future studies of quality can be modified and built, with the downstream goal of an increased understanding of how quality impacts weight of evidence.

A Survey on Reduction of Energy Consumption in Fog Networks-Communications and Computations.

Fog networking has become an established architecture addressing various applications with strict latency, jitter, and bandwidth constraints. Fog Nodes (FNs) allow for flexible and effective computation offloading and content distribution. However, the transmission of computational tasks, the processing of these tasks, and finally sending the results back still incur energy costs. We survey the literature on fog computing, focusing on energy consumption. We take a holistic approach and look at energy consumed by devices located in all network tiers from the things tier through the fog tier to the cloud tier, including communication links between the tiers. Furthermore, fog network modeling is analyzed with particular emphasis on application scenarios and the energy consumed for communication and computation. We perform a detailed analysis of model parameterization, which is crucial for the results presented in the surveyed works. Finally, we survey energy-saving methods, putting them into different classification systems and considering the results presented in the surveyed works. Based on our analysis, we present a classification and comparison of the fog algorithmic models, where energy is spent on communication and computation, and where delay is incurred. We also classify the scenarios examined by the surveyed works with respect to the assumed parameters. Moreover, we systematize methods used to save energy in a fog network. These methods are compared with respect to their scenarios, objectives, constraints, and decision variables. Finally, we discuss future trends in fog networking and how related technologies and economics shall trade their increasing development with energy consumption.

Comparison of Deep Learning and Yolov8 Models for Fox Detection Around the Henhouse

Comparison of Deep Learning and Yolov8 Models for Fox Detection Around the Henhouse

FRACTAL-INSPIRED SENTIMENT ANALYSIS: EVALUATION OF LARGE LANGUAGE MODELS AND DEEP LEARNING METHODS

Sentiment analysis is a vital task in natural language processing (NLP) that aims to identify and extract the emotional states and opinions of text. In this study, we conduct a comprehensive comparison of large language models (LLMs), such as ChatGPT and Google Bard, with conventional methods in sentiment analysis. We employ a rigorous evaluation framework that covers four essential metrics: accuracy, precision, recall, and the [Formula: see text]1-score. Our results reveal that TextBlob outperforms other methods, achieving an impressive accuracy of 69% and precision of 83%. On the other hand, Bard shows a relatively poor performance, with only 39% accuracy and 46% precision. This study offers valuable insights into the diverse capabilities of AI models in sentiment analysis. A key finding of this study is the importance of model selection according to the specific requirements of the task. Each model has its own strengths and weaknesses, which are reflected in their performance profiles. Moreover, the context in which these models operate is crucial. For instance, ChatGPT generates varied responses, Bard struggles with multiple sentences, and Robustly Optimized BERT Pretraining Approach (RoBERTa) balances precision and recall. This study also reveals the performance gap between LLMs and state-of-the-art deep learning methods. We believe this work will inspire future research and applications of ChatGPT and similar AI models in sentiment analysis and related tasks.

Characterization of NK-lysin A, a potent antimicrobial peptide from the zebrafish Danio rerio

Antimicrobial peptides (AMPs) are important players of the innate immune system with a major role in the defense against invading pathogens. AMPs belonging to the family of saposin-like proteins (SAPLIPs) include the porcine NK-lysin and the human granulysin. In the zebrafish Danio rerio, transcript analyses of NK-lysin encoding genes have been reported, but biochemical characterizations at the protein level are missing so far. Here, we present the recombinant expression, purification, and characterization of one of these homologs, namely of NK-lysin A (DaNKlA). To remove the affinity tag from DaNKlA, we made use of a self-splicing intein. Recombinant DaNKlA depolarized liposomes over a broad pH range and showed a preference for negatively charged lipids. DaNKlA inhibited the growth of and killed different Gram-positive and Gram-negative bacteria, including the fish pathogenic bacterium Vibrio anguillarum, by membrane permeabilization but displayed substantially lower activity against yeast cells. Structural modelling and bioinformatic comparison of DaNKlA with characterized SAPLIPs suggest membrane destabilization accompanied by strong electrostatic interactions as the mode of action.

Compositionality and Sentence Meaning: Comparing Semantic Parsing and Transformers on a Challenging Sentence Similarity Dataset

Abstract One of the major outstanding questions in computational semantics is how humans integrate the meaning of individual words into a sentence in a way that enables understanding of complex and novel combinations of words, a phenomenon known as compositionality. Many approaches to modelling the process of compositionality can be classified as either ‘vector-based’ models, in which the meaning of a sentence is represented as a vector of numbers, or ‘syntax-based’ models, in which the meaning of a sentence is represented as a structured tree of labelled components. A major barrier in assessing and comparing these contrasting approaches is the lack of large, relevant datasets for model comparison. This paper aims to address this gap by introducing a new dataset, STS3k, which consists of 2,800 pairs of sentences rated for semantic similarity by human participants. The sentence pairs have been selected to systematically vary different combinations of words, providing a rigorous test and enabling a clearer picture of the comparative strengths and weaknesses of vector-based and syntax-based methods. Our results show that when tested on the new STS3k dataset, state-of-the-art transformers poorly capture the pattern of human semantic similarity judgments, while even simple methods for combining syntax- and vector-based components into a novel hybrid model yield substantial improvements. We further show that this improvement is due to the ability of the hybrid model to replicate human sensitivity to specific changes in sentence structure. Our findings provide evidence for the value of integrating multiple methods to better reflect the way in which humans mentally represent compositional meaning.

Comparison of Explainable AI Models for MRI-based Alzheimer's Disease Classification.

Deep learning models based on convolutional neural networks (CNNs) have been used to classify Alzheimer's disease or infer dementia severity from 3D T1-weighted brain MRI scans. Here, we examine the value of adding occlusion sensitivity analysis (OSA) and gradient-weighted class activation mapping (Grad-CAM) to these models to make the results more interpretable. Much research in this area focuses on specific datasets such as the Alzheimer's Disease Neuroimaging Initiative (ADNI) or National Alzheimer's Coordinating Center (NACC), which assess people of North American, predominantly European ancestry, so we examine how well models trained on these data generalize to a new population dataset from India (NIMHANS cohort). We also evaluate the benefit of using a combined dataset to train the CNN models. Our experiments show feature localization consistent with knowledge of AD from other methods. OSA and Grad-CAM resolve features at different scales to help interpret diagnostic inferences made by CNNs.

A new lightweight framework based on knowledge distillation for reducing the complexity of multi-modal solar irradiance prediction model

The inherent uncertainty of solar energy brings great difficulties to the grid connection and short-term energy planning and dispatching. Deep learning method makes it possible to predict the short-term solar energy with its powerful learning ability, but its complex model structure and huge trainable parameters bring great difficulties to the practical deployment. Therefore, this paper proposes a lightweight framework based on knowledge distillation strategy, which greatly reduces the complexity of multi-modal solar irradiance prediction model meanwhile ensuring an acceptable accuracy, facilitating the practical deployment. Firstly, a teacher model with multi-modal structure and good accuracy is built based on ResNet18-Informer. Then, the lightweight model is obtained by the proposed lightweight framework depending on the knowledge of teacher model. The comparisons of various models and the optimal settings of knowledge distillation are analyzed. Results show that the lightweight model can reduce the trainable parameters, inference time, and GPU memory by 97.7%, 52.5%, and 36.3%, respectively. The normalized root mean square error is reduced by 24.87% compared with the same structure model but without knowledge distillation, verifying the superiority of the proposed framework. The soft loss using the light loss with the ratio of 0.3 can obtain the best training results for the lightweight model. The structure with 3 residual blocks and 3 LSTM layers is proved to be the best for the lightweight model in the solar irradiance prediction task.

Thai Baht and Chinese Yuan Exchange Rate Forecasting Models: ARIMA and SMA-ARIMA Comparison

This study evaluates the effectiveness of classical Autoregressive Integrated Moving Average (ARIMA) models and kth Simple Moving Average - ARIMA (kth SMA-ARIMA) models in forecasting the exchange rate between the Thai Baht (THB) and the Chinese Yuan (CNY). The analysis uses a dataset of historical monthly exchange rates from January 2011 to November 2022, covering 143 months. The dataset is divided into two segments: the initial 127 months are used as the training dataset for model development, while the subsequent 16 months serve as the testing dataset to evaluate forecast accuracy. The Akaike Information Criterion (AIC) is the decision criterion for model selection during the development phase. The forecasting models' effectiveness is subsequently assessed on the testing dataset using two statistical measures: the Mean Absolute Percentage Error (MAPE) and the Root Mean Square Error (RMSE). The findings indicate that the classical ARIMA (0,1,1) model outperforms the kth SMA-ARIMA models in this study, exhibiting the lowest RMSE and MAPE of 0.1702 and 2.6644, respectively. Additionally, a focused comparison of the kth SMA-ARIMA models for k = 2, 3, and 4 reveals that the 2nd SMA-ARIMA (0,1,2) model demonstrates superior performance compared to the 3rd and 4th SMA-ARIMA models. This superiority is reflected in their respective RMSE values of 0.3202, 0.5146, and 0.6339, and corresponding MAPE values of 5.3533, 8.7531, and 10.4949. These results provide valuable insights for decision-makers in the financial sector, enhancing investment strategy formulation based on anticipated currency movements.

A stacking sequence optimization strategy for process‐induced deformation of multi‐layer thick asymmetric laminates

AbstractThe occurrence of process‐induced deformations of composites laminates is challenging for assembly accuracy and may lead to a service life reduction of parts. However, it can be obviously mitigated through different optimization strategies on the basis of the accurate curing process simulation. In this study, a stacking sequence optimization strategy is proposed and applied to multi‐layer thick asymmetric laminates. The shapes of deformed laminate plates are experimentally investigated in virtue of the three‐dimensional coordinate measuring machine. The thermo‐chemical–mechanical behaviors of plates are first verified through the comparisons of model predicted and experimental process‐induced deformations. Then the nonlinear control formula achieved through the regression model is proposed for the direct relationship between stacking sequences and process‐induced deformations. Finally, the required solutions are generated by solving the control formula. With the comparisons between the average deformations before and after optimizations, it is found that the magnitudes of deformations are significantly reduced, especially when the unoptimizated deformations are large.Highlights The thermo‐chemical–mechanical behavior can be described in the curing process simulation model. The deformed shapes of multi‐layer thick asymmetric laminate plates are measured by a three‐dimensional coordinate measuring machine. The stacking sequence optimization strategy is implemented by introducing a nonlinear control formula.

The Lifespan Trajectories of Brain Activities Related to Cognitive Control.

Cognitive control plays a pivotal role in guiding human goal-directed behavior. While existing studies have documented an inverted U-shaped trajectory of cognitive control both behaviorally and anatomically, little is known about the corresponding changes in functional brain activation with age. To bridge this gap, we conducted a comprehensive meta-analysis of 129 neuroimaging studies using conflict tasks, encompassing 3,388 participants aged from 5 to 85 years old. We have three major findings: 1) The inverted U-shaped trajectory is the predominant pattern; 2) Cognitive control-related brain regions exhibit heterogeneous lifespan trajectories: the frontoparietal control network follows inverted U-shaped trajectories, peaking between 24 and 40 years, while the dorsal attention network demonstrates no clear trajectories; 3) Both the youth and the elderly show weaker brain activities and greater left laterality than young to middle-aged adults. These results reveal the lifespan trajectories of cognitive control, highlighting heterogeneous fluctuations in brain networks with age.

Encoding modelling for working memory research: Pattern similarity, representational geometry, and model comparison

Encoding modelling for working memory research: Pattern similarity, representational geometry, and model comparison

Multiparametric MRI-based radiomics combined with 3D deep transfer learning to predict cervical stromal invasion in patients with endometrial carcinoma.

To develop and compare various preoperative cervical stromal invasion (CSI) prediction models, including radiomics, three-dimensional (3D) deep transfer learning (DTL), and integrated models, using single-sequence and multiparametric MRI. Data from 466 early-stage endometrial carcinoma (EC) patients from three centers were collected. Radiomics models were constructed based on T2-weighted imaging (T2WI), diffusion-weighted imaging (DWI), apparent diffusion coefficient (ADC) mapping, contrast-enhanced T1-weighted imaging (CE-T1WI), and four combined sequences as well as 3D DTL models. Two integrated models were created using ensemble and stacking algorithms based on optimal radiomics and DTL models. Model performance and clinical benefits were assessed using area under the curve (AUC), decision curve analysis (DCA), net reclassification index (NRI), integrated discrimination index (IDI), and the Delong test for model comparisons. Multiparametric MRI models were superior to single-sequence models for radiomics or DTL models. Ensemble and stacking integrated models displayed excellent performance. The stacking model had the highest average AUC (0.908) and accuracy (0.883) in external validation groups 1 and 2 (AUC = 0.965 and 0.851, respectively) and emerged as the best predictive model for CSI. All models significantly outperformed the radiologist (P < 0.05). In terms of net benefits, all models demonstrated favorable outcomes in DCA, NRI, and IDI, with the stacking model yielding the highest net benefit. Multiparametric MRI-based radiomics combined with 3D DTL can be used to noninvasively predict CSI in EC patients with greater diagnostic accuracy than the radiologist. Stacking integrated models showed significant potential utility in predicting CSI. Which helps to provide new treatment strategy for clinicians to treat early-stage EC patients.

Development and Comparison of Time Series Models in Predicting Severe Fever with Thrombocytopenia Syndrome Cases - Hubei Province, China, 2013-2020.

Severe fever with thrombocytopenia syndrome (SFTS) is an emerging infectious disease caused by the SFTS virus, which has a high mortality rate. Predicting the number of SFTS cases is essential for early outbreak warning and can offer valuable insights for establishing prevention and control measures. In this study, data on monthly SFTS cases in Hubei Province, China, from 2013 to 2020 were collected. Various time series models based on seasonal auto-regressive integrated moving average (SARIMA), Prophet, eXtreme Gradient Boosting (XGBoost), and long short-term memory (LSTM) were developed using these historical data to predict SFTS cases. The established models were evaluated and compared using mean absolute error (MAE) and root mean squared error (RMSE). Four models were developed and performed well in predicting the trend of SFTS cases. The XGBoost model outperformed the others, yielding the closest fit to the actual case numbers and exhibiting the smallest MAE (2.54) and RMSE (2.89) in capturing the seasonal trend and predicting the monthly number of SFTS cases in Hubei Province. The developed XGBoost model represents a promising and valuable tool for SFTS prediction and early warning in Hubei Province, China.