Lower Prediction Error Research Articles

Evaluate Multi-Objective Optimization Model for Product Supply Chain Inventory Control Based on Grey Wolf Algorithm

Enterprise supply chain inventory control plays a critical role in modern enterprise management. Yet achieving multi-objective optimization remains challenging. This paper proposes a deep learning integrated model based on grey wolf optimization (GWO-DLIM) aimed at optimizing inventory holding costs, stockout costs, and order costs. The model integrates multi-layer perceptron (MLP), DeBERTa, and GWO, leveraging the powerful predictive and feature extraction capabilities of deep learning models, combined with the global optimization ability of the grey wolf algorithm, to achieve multi-objective optimization. These results demonstrate that the GWO-DLIM model significantly outperforms other comparative models across all metrics, exhibiting lower prediction errors and root mean square errors, as well as higher service levels. This study provides an effective multi-objective optimization solution for supply chain inventory management, holding significant theoretical and practical implications.

EXPRESS: A Framework for Execution Time Prediction of Concurrent CNNs on Xilinx DPU Accelerator

Deep learning Processor Unit (DPU) is a highly configurable CNN accelerator that supports a variety of CNNs and can be implemented with multiple instances on the same FPGA. Many applications deploy concurrent execution of different CNNs and in such a setting, an execution time predictor can help “optimize” the DPU configurations to meet the performance requirements of different tasks. We characterize CNN execution on DPUs and reduce the variability in execution time due to interference from the operating system. Subsequently, we propose a machine learning-based framework (EXPRESS) to predict the execution time of any given CNN on a DPU configuration, considering CNN, DPU, and bus characteristics. We improvise EXPRESS to support heterogeneous CNNs in EXPRESS-2.0 by making features independent of the number of CNNs. Our entire experimentation is based on data from a real FPGA board for 16 standard CNNs. Our frameworks, EXPRESS and EXPRESS-2.0, significantly outperform state-of-the-art by achieving an average execution time prediction error of 2.2% and 0.7%, respectively. We illustrate the effectiveness of this low prediction error for design space exploration, which is very useful for embedded system application developers.

IGGCN: Individual-Guided Graph Convolution Network for Pedestrian Trajectory Prediction

Accurately predicting the future trajectory of pedestrians is crucial for applications such as autonomous driving and robot navigation. Graph convolution is widely used in trajectory prediction tasks due to its scalability and adaptive feature-learning capabilities. However, there are two problems with pedestrian trajectory prediction methods based on graph convolution: 1. Previous methods struggled to adjust social interactions according to the attributes of different pedestrians, making it difficult to accurately model the relative importance between different pedestrians and others; 2. Previous methods lacked dynamic processing of pedestrian spatial-temporal interaction features to capture high-level spatial-temporal interaction features effectively. Therefore, we propose an Individual-Guided Graph Convolution Network (IGGCN) for pedestrian trajectory prediction. To tackle problem 1, we design an individual-guided interaction module that can adjust pedestrian social interaction modeling according to the pedestrian's attributes, thereby achieving an accurate description of the relative importance of pedestrians. We extend the module to temporal interaction modeling to further achieve an accurate description of the relative importance of time frames. To address problem 2, we design a deformable convolution module to dynamically process spatial-temporal interaction features through deformable convolution kernels, facilitating the capture of high-level spatial-temporal interaction features. We evaluate our method on the ETH, UCY, and SDD datasets. Quantitative analysis shows that our method has lower prediction errors than the current state-of-the-art methods. Qualitative analysis further reveals that our method effectively eliminates the influence of irrelevant pedestrians and accurately models the spatial-temporal interaction relationship of pedestrians.

Raman spectroscopy coupled with the PLSR model: A rapid method for analyzing gamma-oryzanol content in rice bran oil

Rice bran oil (RBO) is widely used in food, nutraceutical, and cosmetic industries, due to its γ-oryzanol content, a key quality indicator. This study developed a rapid, non-destructive method for quantifying γ-oryzanol in RBO using Raman spectroscopy combined with partial least squares regression (PLSR). The optimal PLSR model, based on orthogonal signal correction (OSC)-pretreated data of Raman spectra from 800 to 1800 cm−1, demonstrated high accuracy with a strong R2-Pearson correlation coefficient of 0.9827 and low root mean square error of prediction (RMSEP) of 0.5314. Principal component analysis (PCA) of OSC-pretreated data showed improved sample grouping by concentration of γ-oryzanol compared to untreated data. Additionally, Bland-Altman plots comparing results from Raman and HPLC methods showed random scatter within ±2 SD of the mean difference, confirming the method's reliability. This study indicates that Raman spectroscopy can serve as a reliable method for determining γ-oryzanol content in RBO products within the related industries.

Volatile Constituents of Cymbopogon citratus (DC.) Stapf Grown in Greenhouse in Serbia: Chemical Analysis and Chemometrics

The present study investigated the volatile constituents of Cymbopogon citratus (lemongrass) grown in a greenhouse environment in Serbia, marking the first commercial cultivation of the plant for essential oil production in the region. The essential oils and hydrolates obtained through steam distillation were analyzed via gas chromatography–mass spectrometry (GC-MS), and the resulting chemical data were further processed using chemometric methods. This study applied quantitative structure retention relationship (QSRR) analysis, employing molecular descriptors (MDs) and artificial neural networks (ANNs) to predict the retention indices (RIs) of the compounds. A genetic algorithm (GA) was used to select the most relevant MDs for this predictive modeling. A total of 29 compounds were annotated in the essential oils, with geranial and neral being the dominant components, while 37 compounds were detected in the hydrolates. The ANN models effectively predicted the RIs of both essential oils and hydrolates, demonstrating high statistical accuracy and low prediction errors. This research offers valuable insights into the chemical profile of lemongrass cultivated in temperate conditions and advances QSRR modeling for essential oil analysis.

A dual-stream temporal convolutional network for remaining useful life prediction of rolling bearings

Abstract Remaining useful life (RUL) prediction plays an indispensable role in the reliable operation and improved maintenance of rolling bearings. Currently, data-driven methods based on deep learning have made significant progress in RUL prediction. However, most of such methods only consider the correlation between channels, ignoring the importance of different time steps for RUL prediction. In addition, it is still challenging to effectively fuse the degradation features of rolling bearings to improve the model RUL prediction performance. To address the above issues, this paper proposes a novel data-driven RUL prediction method named dual-stream temporal convolution network (DSTCN). First, a hybrid attention temporal convolution block (HATCB) is designed to capture the correlation of degraded features on the channel dimension and temporal dimension. Second, a one-dimensional attention fusion module is designed. This module is capable of weight recalibration and assignment to adaptively fuse different degraded features. Afterward, the Hilbert Marginal spectrum is obtained using the Hilbert Huang Transform and used as the input to one stream. Meanwhile, vibration signals are used as the input of the other stream, thus building a dual-stream temporal convolution network to realize RUL prediction. The effectiveness of the proposed method is validated with two life-cycle datasets, and the results show that the method has lower prediction error than other methods for RUL prediction and prognostic analysis.

Exploiting RSM and ANN modeling methods to optimize phosphate ions removal using LDH/alginate composite beads

In the present work, layered double hydroxide (MgAl) and alginate composite beads (LDH/alginate) were developed and used as low-cost and environmentally friendly adsorbent for phosphate ions removal. The successful incorporation of sodium alginate into the LDH structure was confirmed through SEM analysis. The dried beads exhibited a notably rough surface, which promotes molecular movement and potentially enhances pollutant adsorption. The pH of zero charge point (pHPZC) of the composite was found to be 7.41. This result implies that negative ions have a tendency to draw positive charges on the adsorbent surface via electrostatic interaction forces when the pH is lower than the pHPZC. On the other hand, a surface that has a pH higher than pHPZC mostly has a negative charge. The percentage removal and adsorption capacity were investigated as a function of contact time.Experiments were carried out by simultaneously varying three factors, in order to evaluate and compare the predictive capabilities of the response surface methodology (RSM) and the artificial neural network approach (ANN) for the adsorption process. Both methods demonstrated a strong ability to accurately predict the adsorption process. However, the response surface methodology exhibited a lower prediction error compared to the artificial neural network approach.The central composite design within response surface methodology (CCD-RSM) was employed to optimize the experimental conditions for the adsorption process. The model, which considers three factors: adsorbent dose (A), initial concentration (B), and contact time (C), proved to be significant. Among these, the quadratic term (B2) had the most substantial impact on the phosphate ion adsorption rate. The analysis yielded an R2 value of 0.94, indicating an excellent fit to the data. The findings suggest that increasing contact time and reducing the initial concentration improve phosphate ion removal efficiency, while the adsorbent dose has little to no effect. The Artificial Neural Network (ANN) model effectively predicted the adsorptive remediation of phosphate ions onto LDH/alginate composite beads, achieving a high coefficient of determination (R2 = 0.984) between the model outputs and the experimental data. The study highlights the significant potential of LDH/alginate composite beads as a natural adsorbent for the removal of phosphate ions from aqueous solutions. These results underscore the environmental friendliness and efficiency of the adsorbent used for phosphate adsorption.

Advancing photovoltaic panel temperature forecasting: A comparative study of numerical simulation and machine learning in two types of PV power plant

Photovoltaic (PV) panel temperature dynamic monitoring and forecasting is important for managing and maintaining of PV power plant. However, it is uncommon to use a variety of methods to predict and evaluate the panel temperature of different types of PV power plants. Therefore, this study aims to advance PV panel temperature forecasting through a comparative analysis of numerical simulation and machine learning models in two types of PV power plants: land- and water-mounted. The results indicate that PV panel temperature condition for two types of PV power plants can be well captured by the numerical simulation (NS) and machine learning, except for the NS in water-mounted PV power plant (R2 with 0.66). Models perform better in land-mounted PV power plants, with Random Forest Regression (RFR) and ResNet models demonstrating superior accuracy across both environments. While all models exhibit some degree of deviation from normal distribution in residuals, RFR and ResNet show the least deviation and lowest prediction errors, highlighting their robustness in forecasting PV panel temperatures. This study can provide valuable insights for predicting PV panel temperatures and for the management and maintenance of PV power plants.

Multimodal adversarial informer for highway vehicle lane-changing trajectory prediction

To enhance the perception of vehicle trajectory information and lane-changing decision-making capabilities in intelligent connected vehicles during multivehicle interaction scenarios, we propose a novel method based on a Multimodal Adversarial Informer (MAI) for highway multivehicle lane-changingrajectory prediction. This method achieves spatiotemporal features of target and surrounding vehicles through graph learning of temporal features and spatial adjacency matrices. Considering the heading angle and vehicle local X-axis displacement, the vehicle trajectory samples are categorized for training and validation of the multimodal Informer. A multi-criterion discriminator is utilized to judge whether the generated trajectory fits the requirements of accuracy and rationality. After adversarial learning, the optimal vehicle lane-changing trajectory prediction is obtained using the proposed MAI. Experiments conducted with the NGSIM dataset demonstrate the comparative performance of baseline models on three different noise-added testing datasets using MAE, RMSE, and R² metrics. The MAI model consistently outperforms the others, achieving the lowest MAE and RMSE and the highest R² values across all datasets, indicating superior predictive accuracy and fit. Furthermore, the results show that the proposed MAI framework maintains a relatively low prediction error over both short-term and long-term horizons compared to baseline models.

Analyzing Machine Learning Models for Activity Recognition Using Homomorphically Encrypted Real-World Smart Home Datasets: A Case Study

The era of digitization and IoT devices is marked by the constant storage of massive amounts of data. The growing adoption of smart home environments, which use sensors and devices to monitor and control various aspects of daily life, underscores the need for effective privacy and security measures. HE is a technology that enables computations on encrypted data, preserving confidentiality. As a result, researchers have developed methodologies to protect user information, and HE is one of the technologies that make it possible to perform computations directly on encrypted data and produce results using this encrypted information. Thus, this research study compares the performance of three ML models, XGBoost, Random Forest, and Decision Classifier, on a real-world smart home dataset using both with and without FHE. Practical results demonstrate that the Decision Classifier showed remarkable results, maintaining high accuracy with FHE and even surpassing its plaintext performance, suggesting that encryption can enhance model accuracy under certain conditions. Additionally, Random Forest showed efficiency in terms of execution time and low prediction errors with FHE, making it a strong candidate for encrypted data processing in smart homes. These findings highlight the potential of FHE to set new privacy standards, advancing secure and privacy-preserving technologies in smart environments.

The risk factors determined by four machine learning methods for the change of difference of bone mineral density in post-menopausal women after three years follow-up

The prevalence of osteoporosis has drastically increased recently. It is not only the most frequent but is also a major global public health problem due to its high morbidity. There are many risk factors associated with osteoporosis were identified. However, most studies have used the traditional multiple linear regression (MLR) to explore their relationships. Recently, machine learning (Mach-L) has become a new modality for data analysis because it enables machine to learn from past data or experiences without being explicitly programmed and could capture nonlinear relationships better. These methods have the potential to outperform conventional MLR in disease prediction. In the present study, we enrolled a Chinese post-menopause cohort followed up for 4 years. The difference of T-score (δ-T score) was the dependent variable. Information such as demographic, biochemistry and life styles were the independent variables. Our goals were: (1) Compare the prediction accuracy between Mach-L and traditional MLR for δ-T score. (2) Rank the importance of risk factors (independent variables) for prediction of δ T-score. Totally, there were 1698 postmenopausal women were enrolled from MJ Health Database. Four different Mach-L methods namely, Random forest (RF), eXtreme Gradient Boosting (XGBoost), Naïve Bayes (NB), and stochastic gradient boosting (SGB), to construct predictive models for predicting δ-BMD after four years follow-up. The dataset was then randomly divided into an 80% training dataset for model building and a 20% testing dataset for model testing. A 10-fold cross-validation technique for hyperparameter tuning was used. The model with the lowest root mean square error for the validation dataset was viewed as the best model for each ML method. The averaged metrics of the RF, SGB, NB, and XGBoost models were used to compare the model performance of the benchmark MLR model that used the same training and testing dataset as the Mach-L methods. We defined that the priority demonstrated in each model ranked 1 as the most critical risk factor and 22 as the last selected risk factor. For Pearson correlation, age, education, BMI, HDL-C, and TSH were positively and plasma calcium level, and baseline T-score were negatively correlated with δ-T score. All four Mach-L methods yielded lower prediction errors than the MLR method and were all convincing Mach-L models. From our results, it could be noted that education level is the most important factor for δ-T Score, followed by DBP, smoking, SBP, UA, age, and LDL-C. All four Mach-L outperformed traditional MLR. By using Mach-L, the most important six risk factors were selected which are, from the most important to the least: DBP, SBP, UA, education level, TG and sleeping hour. δ T score was positively related to SBP, education level, UA and TG and negatively related to DBP and sleeping hour in postmenopausal Chinese women.

Comparison of the Accuracy Between the Z CALC2 Calculator and Barrett Toric Calculator in Toric IOL Calculation.

To compare the accuracy of the Z CALC2 calculator and Barrett toric calculator in toric intraocular lens (IOL) calculation. Eighty-five eyes of 85 patients who underwent uneventful cataract surgery with toric IOL implantation were included. The accuracy was compared between the Z CALC2 calculator and Barrett toric calculator under two calculation modes: using simulated keratometry (SimK) from the IOLMaster 700 (Carl Zeiss Meditec AG) for posterior corneal astigmatism (PCA) prediction and employing total corneal astigmatism (total corneal refractive power [TCRP] or measured PCA) obtained from the Pentacam (Oculus Optikgeräte GmbH). The centroid of prediction errors, percentage of eyes with prediction errors within ±0.50 diopter (D), mean prediction error, and mean absolute prediction error were calculated. Subgroup analysis was conducted based on the orientation and magnitude of anterior corneal astigmatism and axial length. When using SimK, the two calculators with predicted PCA showed comparable accuracy. When employing total corneal astigmatism, the Barrett toric calculator with measured PCA showed a lower centroid error (0.15 vs 0.38 D), a higher percentage of eyes with prediction errors within ±0.50 D (47.1% vs 32.9%, P = .018), and a lower mean prediction error (0.57 vs 0.71 D, P = .033) compared to the Z CALC2 calculator with TCRP in the 4-mm zone. In subgroup analysis, when employing total corneal astigmatism, the Barrett toric calculator with measured PCA exhibited superior accuracy, especially in the with-the-rule and anterior corneal astigmatism of 2.00 D or less subgroups. When using SimK, the Z CALC2 calculator and Barrett toric calculator yield comparable accuracy. The Barrett toric calculator with measured PCA may be more recommended when employing total corneal astigmatism. [J Refract Surg. 2024;40(10):e681-e691.].

IMPLEMENTASI MACHINE LEARNING ALGORITMA NEURAL NETWORK DALAM MEMPREDIKSI HARGA SAHAM MENGGUNAKAN RAPIDMINER

The research is aimed at implementing the Neural Network algorithm in predicting stock prices using RapidMiner. Historical stock price data is analyzed to train neural network models in identifying patterns and trends that can be used to predict future stock prices. The research phases include data collection, data preprocessing, Neural Network model development, model training, and model performance evaluation. Data used include daily closing prices, trade volumes, and other technical indicators. The data is processed through several stages of preprocesing, including normalization and filling of missing values, before being used to train Neural network models. The model was built with several layers and the number of neurons adjusted based on early experiments to find the optimal configuration. Models are trained using datasets that are divided into training sets and testing sets. The predicted result is compared to the actual value to evaluate the performance of the model. The results showed that the Neural Network model was able to predict stock prices with a fairly high accuracy rate, with an RMSE of 0.989 indicating a low prediction error rate. The research concludes that the Neural Network algorithm implemented in RapidMiner is effective in predicting stock prices and has significant potential to be used in investment decision-making. The use of RapidMiner as a data processing tool has proven to be efficient and facilitating the development and evaluation of predictive models. Keywords: Neural Network, Stock Price Prediction, RapidMiner, Machine Learning, Investment

Chemometric approaches for polysaccharide viscosity profiling

Traditional viscosity measurements for carrageenan are laborious, present practical and environmental challenges, and fail to provide structure-property understanding for application and manufacturing development. We hypothesize that integrating Size Exclusion Chromatography (SEC) with Multi-Angle Light Scattering (MALS) and online viscometry, combined with chemometric techniques, can develop a more efficient and environmentally friendly method for determining the apparent viscosity of carrageenan solutions.To test this hypothesis, predictive chemometric models were developed using SEC-MALS data for carrageenan extracted from four different seaweed species. By integrating SEC-MALS with Partial Least Squares (PLS) regression, key molecular parameters such as hydrodynamic radius, intrinsic viscosity, and molecular mass were identified as significant influencers of viscosity. The model for carrageenan from Eucheuma denticulatum yielded the lowest prediction error (RMSEP 8.4), while those for carrageenan extracted from Kappaphycus alvarezii or from several species of the Chondrus genus showed higher errors due to κ-carrageenan sensitivity. For carrageenan extracted from seaweed of the Gigartina genus, incorporating the root mean square radius resulted in a low prediction error of 10.This study concludes that integrating SEC-MALS with PLS regression effectively identifies key molecular parameters influencing carrageenan viscosity, enhancing structure-property understanding and providing a reliable analytical method for optimizing quality control and application in various industries.

Machine Learning-Empowered Real-Time Acoustic Trapping: An Enabling Technique for Increasing MRI-Guided Microbubble Accumulation

Acoustic trap, using ultrasound interference to ensnare bioparticles, has emerged as a versatile tool for life sciences due to its non-invasive nature. Bolstered by magnetic resonance imaging’s advances in sensing acoustic interference and tracking drug carriers (e.g., microbubble), acoustic trap holds promise for increasing MRI-guided microbubbles (MBs) accumulation in target microvessels, improving drug carrier concentration. However, accurate trap generation remains challenging due to complex ultrasound propagation in tissues. Moreover, the MBs’ short lifetime demands high computation efficiency for trap position adjustments based on real-time MRI-guided carrier monitoring. To this end, we propose a machine learning-based model to modulate the transducer array. Our model delivers accurate prediction of both time-of-flight (ToF) and pressure amplitude, achieving low average prediction errors for ToF (−0.45 µs to 0.67 µs, with only a few isolated outliers) and amplitude (−0.34% to 1.75%). Compared with the existing methods, our model enables rapid prediction (<10 ms), achieving a four-order of magnitude improvement in computational efficiency. Validation results based on different transducer sizes and penetration depths support the model’s adaptability and potential for future ultrasound treatments.

Development of Machine Learning and Deep Learning Prediction Models for PM2.5 in Ho Chi Minh City, Vietnam

The application of machine learning and deep learning in air pollution management is becoming increasingly crucial, as these technologies enhance the accuracy of pollution prediction models, facilitating timely interventions and policy adjustments. They also facilitate the analysis of large datasets to identify pollution sources and trends, ultimately contributing to more effective and targeted environmental protection strategies. Ho Chi Minh City (HCMC), a major metropolitan area in southern Vietnam, has experienced a significant rise in air pollution levels, particularly PM2.5, in recent years, creating substantial risks to both public health and the environment. Given the challenges posed by air quality issues, it is essential to develop robust methodologies for predicting PM2.5 concentrations in HCMC. This study seeks to develop and evaluate multiple machine learning and deep learning models for predicting PM2.5 concentrations in HCMC, Vietnam, utilizing PM2.5 and meteorological data over 911 days, from 1 January 2021 to 30 June 2023. Six algorithms were applied: random forest (RF), extreme gradient boosting (XGB), support vector regression (SVR), artificial neural network (ANN), generalized regression neural network (GRNN), and convolutional neural network (CNN). The results indicated that the ANN is the most effective algorithm for predicting PM2.5 concentrations, with an index of agreement (IOA) value of 0.736 and the lowest prediction errors during the testing phase. These findings imply that the ANN algorithm could serve as an effective tool for predicting PM2.5 concentrations in urban environments, particularly in HCMC. This study provides valuable insights into the factors that affect PM2.5 concentrations in HCMC and emphasizes the capacity of AI methodologies in reducing atmospheric pollution. Additionally, it offers valuable insights for policymakers and health officials to implement targeted interventions aimed at reducing air pollution and improving public health.

Developing a digital mapping of soil organic carbon on a national scale using Sentinel-2 and hybrid models at varying spatial resolutions

Mapping the spatial distribution of soil organic carbon (SOC) is crucial for monitoring soil health, understanding ecosystem functions, and contributing to global carbon cycling. However, few studies have directly compared the influence of hybrid models and individual models with varying spatial resolutions on SOC prediction at a national scale. In this study, by combining remote sensing data, we utilized the LUCAS 2018 soil dataset to evaluate the potential capacities of hybrid models for predicting SOC content at different spatial resolutions in Germany. The hybrid models PLSRK and RFK consisted of partial least square regression (PLSR) with residual original kriging (OK) models, and random forest (RF) models with residual OK models, respectively. Individual PLSR and RF models were used as reference models. All these models were applied to estimate SOC content at 10 m, 50 m, 100 m, and 200 m spatial resolutions. Sentinel-2 bands, band indices, and topography variables were as predictors. The results revealed that hybrid models had a more accurate prediction of SOC content with higher explanations and lower prediction errors compared with individual models. The RFK model at the spatial resolution of 100 m was the fittest model with R2 = 0.416, RMSE = 0.545, and RPIQ = 1.647, which enhanced 3.74% of explanation compared with the performance of RF model. The results also showed that hybrid models at a relatively coarse resolution (100 m) had better accuracy instead of those at high spatial resolution (10 m, 50 m). Sentinel-2 remote sensing data showed significant predictive capabilities for estimating SOC content. The predicted spatial distribution of SOC content revealed that the high SOC concentrated in the northwest grassland, central and southwestern mountains, and the Alps in Germany. Our study provided a benchmark SOC map in Germany for monitoring the changes resulting from land use and climate impacts, and we illustrated the accuracy of hybrid models and the effects of spatial resolutions on SOC predictions at a national scale.

Gap filling of missing satellite data from MODIS and CMEMS for chlorophyll-a in the waters of Aceh, Indonesia

The motivation behind our study is to identify a robust method to enhance the accuracy of missing data, particularly chlorophyll-a data, which often goes undetected due to various factors. This study analyzes chlorophyll-a concentrations and sea level changes due to tides using three methods: Linear Interpolation, Fillgaps, and Modified Fillgaps. Two experiments were conducted: Experiment I involved random data removal (60% and 70%), and Experiment II combined sequential and random data removal (25% sequentially on the right, 35% and 45% randomly on the left). In Experiment I, the Modified Fillgaps method showed high correlation coefficients (up to 0.96) between original and reconstructed data, demonstrating its effectiveness in accurately filling significant data gaps. This method also exhibited low Root Mean Square Error and Mean Absolute Error values, confirming its predictive precision. In Experiment II, despite structured and realistic data loss patterns, the method maintained high correlation and low prediction errors, with low Normalized Root Mean Squared Error and Mean Absolute Percentage Error values, further validating its reliability. Additionally, the method excelled in two-dimensional chlorophyll-a maps, outperforming Linear Interpolation and Fillgaps methods in scenarios with 50% and 60% data loss, achieving higher correlation and lower prediction errors. These findings are crucial for environmental and climatological studies relying on satellite-derived data, confirming the Modified Fillgaps method as the most reliable and effective for handling significant data loss in chlorophyll-a map analyses. Future research should explore its application to other environmental data types and more complex data loss patterns.

Real-time prediction of metacentric height of ro-ro passenger ships in Qiongzhou strait based on improved RBF neural network

Addressing the complexities and real-time challenges in calculating ship metacentric height (GM), this study proposes an improved method using an optimized radial basis function neural network (RBFNN) for real-time GM prediction. Bayesian optimization is introduced to fine-tune the hyperparameters of the RBFNN, aiming to enhance the model's generalization performance. The study focuses on the Qiongzhou Strait Ro-ro passenger ship ‘Zijing No.11’ and selects three GM-related factors as neural network inputs using grey correlation analysis. The GM calculated by the empirical formula serves as the expected value, which is compared with predictions from various algorithms. Simulation results demonstrate that the improved RBFNN achieves significantly lower prediction errors compared to the unoptimized version. Furthermore, compared to other machine learning models and artificial neural networks, the proposed model exhibits superior performance in predicting ship initial stability height. Consequently, this model offers a practical tool for accurate and real-time GM prediction, enhancing intelligent stowage and operational efficiency in shipping.

Utilizing textual data from online reviews for daily tourism demand forecasting: A deep learning approach leveraging word embedding techniques

Accurately estimating daily tourism volumes is crucial for optimizing operational strategies and enhancing visitor experiences at tourist destinations. In this study, we leverage historical tourism volume data, search engine data, and online review data (including textual review contents) to forecast daily tourism demand. We develop a deep learning framework that includes a feature selection module to select search engine indices, a word embedding module to transform review texts into numerical predictors, and an extreme learning machine (ELM) enhanced with the whale optimization algorithm (WOA) for predictive modeling. Based on different word embedding techniques, we investigated two specific forecasting methods: one based on term frequency-inverse document frequency (TF-IDF) and Transformer, and the other based on the standard pre-trained bidirectional encoder representations from Transformers (BERT). These two methods enhance predictive accuracy while ensuring fast training and inference speeds, making it suitable for the high-frequency daily tourism forecasting task. Experimental results demonstrate that the two methods significantly outperform traditional approaches that rely solely on numerical data sources, achieving lower prediction errors and faster inference speeds compared to established benchmark methods, highlighting their potential to contribute to advancements in tourism prediction methodologies.