Low Error Research Articles

Explainable Pre-Trained Language Models for Sentiment Analysis in Low-Resourced Languages

Sentiment analysis is a crucial tool for measuring public opinion and understanding human communication across digital social media platforms. However, due to linguistic complexities and limited data or computational resources, it is under-represented in many African languages. While state-of-the-art Afrocentric pre-trained language models (PLMs) have been developed for various natural language processing (NLP) tasks, their applications in eXplainable Artificial Intelligence (XAI) remain largely unexplored. In this study, we propose a novel approach that combines Afrocentric PLMs with XAI techniques for sentiment analysis. We demonstrate the effectiveness of incorporating attention mechanisms and visualization techniques in improving the transparency, trustworthiness, and decision-making capabilities of transformer-based models when making sentiment predictions. To validate our approach, we employ the SAfriSenti corpus, a multilingual sentiment dataset for South African under-resourced languages, and perform a series of sentiment analysis experiments. These experiments enable comprehensive evaluations, comparing the performance of Afrocentric models against mainstream PLMs. Our results show that the Afro-XLMR model outperforms all other models, achieving an average F1-score of 71.04% across five tested languages, and the lowest error rate among the evaluated models. Additionally, we enhance the interpretability and explainability of the Afro-XLMR model using Local Interpretable Model-Agnostic Explanations (LIME) and Shapley Additive Explanations (SHAP). These XAI techniques ensure that sentiment predictions are not only accurate and interpretable but also understandable, fostering trust and reliability in AI-driven NLP technologies, particularly in the context of African languages.

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.

Enhancing cognitive abilities in young adults with ADHD through instrumental music training: a comparative analysis of musicians and non-musicians.

Extensive research highlights the multifaceted benefits of active musical engagement across all ages, from childhood to the elderly. The practice of a musical instrument activates numerous brain regions, enhancing a range of neurocognitive skills. Despite accumulating evidence from various clinical populations, research on the effects of musical training in individuals with ADHD is scarce, with virtually no studies focusing on adults. This study aims to fill the gap by evaluating the impact of long-term instrumental music training on cognitive abilities in young adults (18-35 years) diagnosed with ADHD. Cognitive abilities were compared across groups of 48 musicians (experienced guitar or piano players) and 46 matched non-musicians, all confirmed to have ADHD. The assessments covered cognitive domains such as sustained attention, visuospatial processing, processing speed, graphomotor speed, working memory, auditory recall, response inhibition, and executive function. Evaluation tools included the Digit-Symbol Coding Test, Digit Span Test, Symbol Search Test, Switching Task, and Continuous Performance Test (CPT). Collectively, the results indicated a notable enhancement in cognitive performance in the musician group compared to the non-musician group, including in functions central to the disorder, such as sustained attention and impulse control. Musicians scored higher on the Digit-Symbol Coding, Digit Span, and Symbol Search tests, showed lower error rates and greater consistency in reaction times in the Switching Task, and had fewer commission errors in the CPT. The findings support the integration of specialized musical training in therapeutic and support programs for ADHD, suggesting benefits that may extend into adulthood.

NitroNet – a machine learning model for the prediction of tropospheric NO2 profiles from TROPOMI observations

Abstract. We introduce NitroNet, a deep learning model for the prediction of tropospheric NO2 profiles from satellite column measurements. NitroNet is a neural network trained on synthetic NO2 profiles from the regional chemistry and transport model WRF-Chem, which was operated on a European domain for the month of May 2019. This WRF-Chem simulation was constrained by in situ and satellite measurements, which were used to optimize important simulation parameters (e.g. the boundary layer scheme). The NitroNet model receives NO2 vertical column densities (VCDs) from the TROPOspheric Monitoring Instrument (TROPOMI) and ancillary variables (meteorology, emissions, etc.) as input, from which it reproduces NO2 concentration profiles. Training of the neural network is conducted on a filtered dataset, meaning that NO2 profiles showing strong disagreement (>20 %) with colocated TROPOMI column measurements are discarded. We present a first evaluation of NitroNet over a variety of geographical and temporal domains (Europe, the US West Coast, India, and China) and different seasons. For this purpose, we validate the NO2 profiles predicted by NitroNet against satellite, in situ, and MAX-DOAS (Multi-Axis Differential Optical Absorption Spectroscopy) measurements. The training data were previously validated against the same datasets. During summertime, NitroNet shows small biases and strong correlations with all three datasets: a bias of +6.7 % and R=0.95 for TROPOMI NO2 VCDs, a bias of −10.5 % and R=0.75 for AirBase surface concentrations, and a bias of −34.3 % to +99.6 % with R=0.83–0.99 for MAX-DOAS measurements. In comparison to TROPOMI satellite data, NitroNet even shows significantly lower errors and stronger correlation than a direct comparison with WRF-Chem numerical results. During wintertime considerable low biases arise because the summertime/late-spring training data are not fully representative of all atmospheric wintertime characteristics (e.g. longer NO2 lifetimes). Nonetheless, the wintertime performance of NitroNet is surprisingly good and comparable to that of classic regional chemistry and transport models. NitroNet can demonstrably be used outside the geographic and temporal domain of the training data with only slight performance reductions. What makes NitroNet unique when compared to similar existing deep learning models is the inclusion of synthetic model data, which offers important benefits: due to the lack of NO2 profile measurements, models trained on empirical datasets are limited to the prediction of surface concentrations learned from in situ measurements. NitroNet, however, can predict full tropospheric NO2 profiles. Furthermore, in situ measurements of NO2 are known to suffer from biases, often larger than +20 %, due to cross-sensitivities to photooxidants, which other models trained on empirical data inevitably reproduce.

Navigating space: how fine and gross motor expertise influence spatial abilities at different scales.

Spatial ability, essential for navigating and interacting with the environment, comprises small-scale (e.g., mental rotation) and large-scale (e.g., spatial navigation) skills. Previous research underscores the influence of motor expertise on these abilities, yet comparative studies among different types of movement experts are limited, especially regarding the impact of gross motor skills on large-scale spatial abilities. This case-control study compared small-scale and large-scale spatial abilities among fine movement experts, gross movement experts, and non-movement experts. Ninety participants (30 per group) were assessed through computer-based spatial ability tests, including the Revised Purdue Spatial Visualization Test (PSVT: R), Mental Rotation Test, a navigation task developed in Unity 3D, and Triangle Completion Test (TCT). Fine movement experts excelled in small-scale spatial tasks compared to non-movement experts. Gross movement experts demonstrated superior large-scale spatial abilities, evidenced by lower errors in TCT and higher navigation scores, distinguishing their performance in spatial navigation and orientation from both fine movement experts and non-movement experts. The study highlights the distinct impacts of fine and gross motor expertise on spatial abilities, with gross motor skills particularly benefiting large-scale spatial navigation. These findings suggest potential clinical applications of gross motor training for improving spatial abilities in neurological populations, advocating for further research in immersive virtual environments and exploring lateral dominance effects on spatial performance.

Hybrid Machine Learning Approaches for 5G Traffic Prediction

ABSTRACT Accurate traffic prediction poses great difficulties because of the continuously increasing scale and diversity of 5G network traffic, which is driven by user demands. Moreover, certain characteristics of 5G traffic are constantly changing; thus, simulations using traditional models often lead to incorrect estimations or inefficient utilization of available resources. Consequently, we propose a hybrid machine learning model that integrates support vector machine (SVM) and decision tree algorithms to enhance efficiency of 5G traffic prediction. The structure of the hybrid model dynamically adjusts by adding or removing hidden layers and units within the network to improve prediction performance. The efficacy of the proposed model is evaluated using metrics like mean squared error, mean absolute error, and root mean squared error (RMSE). Findings show that the hybrid model consistently achieves lower error rates than SVM alone. Further performance enhancement of the hybrid model in predicting 5G traffic is also supported by comparisons of R-squared values against signal-to-noise ratios. These outcomes show the potential of the proposed method to improve traffic prediction accuracy in 5G networks, serving as a powerful tool for network control.

Low-overhead quantum computing with the color code

Fault-tolerant quantum computation demands significant resources: large numbers of physical qubits must be checked for errors repeatedly to protect quantum data as logic gates are implemented in the presence of noise. We demonstrate that an approach based on the color code can lead to considerable reductions in the resource overheads compared with conventional methods, while remaining compatible with a two-dimensional layout. We propose a lattice surgery scheme that exploits the rich structure of the color-code phase to perform arbitrary pairs of commuting logical Pauli measurements in parallel while keeping the space cost low. Compared to lattice surgery schemes based on the surface code with the same code distance, and assuming the same amount of time is needed to complete a round of syndrome measurements, our approach yields about a 3× improvement in the space-time overhead, obtained from a combination of a 1.5× improvement in spatial overhead together with a 2× speedup due to the parallelization of commuting logical measurements. Even when taking into account the color code's lower error threshold using current decoders, the overhead is reduced by 10% at a physical error rate of 10−3 and by 50% at 10−4. Published by the American Physical Society 2024

Enhancing positioning accuracy in adjuvant radiotherapy for left breast cancer using cervical-thoracic integrated bracket combined with deep inspiration breath holding.

This study aimed to investigate the accuracy of three fixation methods in patients with left breast cancer receiving whole breast radiotherapy: conventional breast bracket (BB), breast bracket combined with deep inspiration breath holding (DIBH), and cervical-thoracic integrated bracket (CTIB) combined with DIBH. From January 2023 to September 2023, 84 patients who underwent left breast cancer radiotherapy with supraclavicular radiation after conservative surgery were included in this study, of which 25 patients were fixed by conventional BB, 34 patients by BB & DIBH, and 25 patients by CTIB & DIBH. Image registration was conducted around the treatment area, using the sternoclavicular joint and acromioclavicular joint as landmarks. Systematic and random errors were calculated to assess the accuracy of these fixation methods. Compared to the conventional BB group, the CTIB & DIBH group demonstrated significant improvements in accuracy across multiple dimensions, including left-right, superior-posterior, and anterior-posterior directions, as well as rotational errors in the sagittal and coronal planes. The CTIB & DIBH group showed a significant reduction of setup error in the anterior-posterior direction compared to the BB & DIBH group. The displacement of the acromioclavicular joint varied, with the CTIB & DIBH method showing more favorable outcomes. DIBH method exhibited lower setup errors and more effective fixation of the acromioclavicular joint, especially when combined with CTIB, making it a recommended fixation method in adjuvant radiotherapy following breast-conserving surgery.

Predicting COD and TN in A2O+AO Process Considering Influent and Reactor Variability: A Dynamic Ensemble Model Approach

The prediction of the chemical oxygen demand (COD) and total nitrogen (TN) in integrated anaerobic–anoxic–oxic (A2O) and anoxic–oxic (AO) processes (i.e., A2O+AO process) was achieved using a dynamic ensemble model that reflects the dynamics of wastewater treatment plants (WWTPs). This model effectively captures the variability in the influent characteristics and fluctuations within each reactor of the A2O+AO process. By employing a time-lag approach based on the hydraulic retention time (HRT), artificial intelligence (AI) selects suitable input (i.e., pH, temperature, total dissolved solid (TDS), NH3-N, and NO3-N) and output (COD and TN) data pairs for training, minimizing the error between predicted and observed values. Data collected over two years from the actual A2O+AO process were utilized. The ensemble model adopted machine learning-based XGBoost for COD and TN predictions. The dynamic ensemble model outperformed the static ensemble model, with the mean absolute percentage error (MAPE) for the COD ranging from 9.5% to 15.2%, compared to the static ensemble model’s range of 11.4% to 16.9%. For the TN, the dynamic model’s errors ranged from 9.4% to 15.5%, while the static model showed lower errors in specific reactors, particularly in the anoxic and oxic stages due to their stable characteristics. These results indicate that the dynamic ensemble model is suitable for predicting water quality in WWTPs, especially as variability may increase due to external environmental factors in the future.

Lifetime mapping using femtosecond time-resolved photoemission electron microscopy.

Time-resolved photoemission electron microscopy (PEEM) has established itself as a versatile experimental technique to unravel the ultrafast electron dynamics of materials with nanometer-scale resolution. However, the approach of performing PEEM-based, pixel-by-pixel lifetime mapping has not been reported thus far. Herein, we describe in detail the data pre-processing procedure and an algorithm to perform time-trace fittings of each pixel. We impose an energy cutoff for each pixel prior to spectral integration to enhance the robustness of our approach. With the energy cutoff, the energy-integrated time traces show improved statistics and lower fitting errors, thus resulting in a more accurate determination of the fit parameters, e.g., decay time constants. Our work allows us to reliably construct PEEM-based lifetime maps, which potentially shed light on the effects of local microenvironment on the ultrafast processes of the material and allow spatial distributions of lifetimes to be correlated with observables obtained from complementary microscopic techniques, hence enabling a more comprehensive characterization of the material.

A Picture is Worth a Thousand Timesteps: Excess Entropy Scaling for Rapid Estimation of Diffusion Coefficients in Molecular-Dynamics Simulations of Fluids.

In molecular-dynamics simulations of fluids, the Einstein-Helfand (EH) and Green-Kubo (GK) relationships are frequently used to compute a variety of transport coefficients, including diffusion coefficients. These relationships are formally valid in the limit of infinite sampling time: The error in the estimate of a transport coefficient (relative to an infinitely long simulation) asymptotically approaches zero as more dynamics are simulated and recorded. In practice, of course, one can only simulate a finite number of particles for a finite amount of time. In this work, we show that in this pre-asymptotic regime, an approach for estimating diffusion coefficients based upon excess entropy scaling (EES) achieves a significantly lower error than either EH or GK relationships at fixed online sampling time. This approach requires access only to structural information at the level of the radial distribution function (RDF). We further demonstrate that the use of a recently developed RDF mollification scheme significantly reduces the amount of sampling time needed to converge to the long-time value of the diffusion coefficient. We also demonstrate favorable sample-to-sample variances in the diffusion coefficient estimate obtained using EES as compared to those obtained using EH and GK.

Chitosan impregnated sugarcane bagasse biochar for removal of anionic dyes from wastewater

Wastewater treatment is of utmost importance in providing all equitable and safe drinking water. In the present study, a chitosan impregnated sugarcane bagasse biochar SCNC biocomposite has been synthesized for the removal of Congo red (CR) dye from an aqueous solution. The SCNC biocomposite was thoroughly characterized through Brunauer, Emmett and Teller (BET) and N2 adsorption isotherm, point of zero charge (pHPZC), elemental analysis, Scanning electron microscopy (SEM), Fourier transform infrared spectroscopy (FTIR) and Thermogravimetric analysis (TGA) analysis. Moreover, SCNC biocomposite was further employed to remove CR dye from the aqueous solution in batch mode. The SCNC biocomposite could remove more than 95.0% of CR at an initial concentration of (100mgL− 1), adsorbent dosage (0.05 g), time (200 min), pH ~ 3. The SCNC biocomposite achieved maximum adsorption capacity of 170mgg− 1. The equilibrium adsorption data for CR dye were best fitted to the Langmuir isotherm model with R2, 0.999. The kinetic and isotherm were statistically investigated using the chi-square statistic (χ2 ), mean square error (MSE), and the sum of squares error (SSE) Because of the higher correlation coefficient (R2 ≥ 0.999) and lower error functions, the equilibrium CR adsorption isotherms for a single-dye system fit Langmuir and the PSO kinetic model. The thermodynamic studies revealed the spontaneous and endothermic nature of adsorption of CR dye onto SCNC biocomposite. The SCNC biocomposite can be regenerated up to the 5th cycle successfully. The mechanism of CR adsorption onto SCNC was elucidated.

Accurate Acupoint Localization in 2D Hand Images: Evaluating HRNet and ResNet Architectures for Enhanced Detection Performance

Introduction: This research assesses HRNet and ResNet architectures for their precision in localizing hand acupoints on 2D images, which is integral to automated acupuncture therapy. Objectives: The primary objective was to advance the accuracy of acupoint detection in traditional Korean medicine through the application of these advanced deep-learning models, aiming to improve treatment efficacy. Background: Acupoint localization in traditional Korean medicine is crucial for effective treatment, and the study aims to enhance this process using advanced deep-learning models. Methods: The study employs YOLOv3, YOLOF, and YOLOX-s for object detection within a top-down framework, comparing HRNet and ResNet architectures. These models were trained and tested using datasets annotated by technicians and their mean values, with performance evaluated based on Average Precision at two IoU thresholds. Results: HRNet consistently demonstrated lower mean distance errors across various acupoints compared to ResNet, particularly at a 256x256 pixel resolution. Notably, the HRNet-w48 model surpassed human annotators, including medical experts, in localization accuracy. Conclusion: HRNet's superior performance in acupoint localization suggests its potential to improve the precision and efficacy of acupuncture treatments. The study highlights the promising role of machine learning in enhancing traditional medical practices and underscores the importance of accurate acupoint localization in clinical acupuncture.

Two-step fabrication of clear view SLA millifluidic device for long-term in-vitro cultures

Purpose In vitro millifluidic cultures with perfusion are essential tools to analyse and understand the interactions between cells, their matrix and multi-cell populations. The purpose of this paper is to focus on the design and development of a 3D-printed template that can be used for fabrication of a clear view poly (dimethyl siloxane) (PDMS) device. The major objective is to obtain a transparent device prototype that allows perfusion culture of two cell types for multiple days that can be imaged using laser scanning confocal microscopy. Design/methodology/approach The authors used a two-step approach for achieving the final geometric structure at a faster timeline and lower cost. The first part focuses on comparing the fidelity of the printing templates using fused deposition modelling (FDM) and stereolithography (SLA) printers for a range of dimensions. They then show that the complex geometry chip with connection chambers can be printed using low resolution low cost FormLab SLA printer. The final optimized design was then printed using high-resolution Projet 6000 SLA printer to obtain smoother structures. Findings In this work, the authors have shown that the FormLab SLA printer yields significantly lower error for printing complex design geometries as compared to FDM printer. Result shows that FormLab printer can be used to achieve a minimum dimension of 0.5 mm. They then use the printer to optimize the device dimension for the culture chip which requires several iterations of printing and experimenting. They showed the two-step protocol of printing the optimized template in a high-resolution SLA printer and further fabricating a clear view millifluidic PDMS device that is compatible confocal microscopy imaging. They used this culture chip for perfusion culture of two cell type, and the controlled fluidic exchange between the two chambers led to the formation of neuroglia junction. Originality/value One of the major bottlenecks for obtaining complex geometry in mili/microfluidic device by 3D printing is the need of multiple iterations on printing. This makes the tuning of dimension significantly expensive. Another challenge is to obtain a smooth surface of PDMS that leads to a leak proof clear view device compatible for laser based confocal imaging. The combination of two printers plays a crucial role for the rapid prototyping of the imaging device with flow control. The proposed approach lowers the cost for prototyping of in vitro culture chip with complex geometries to improve on biological research demanding multi-chamber fluidic device.

Estimating cognitive workload using a commercial in-ear EEG headset.

This study investigated the potential of estimating various mental workload levels during two different tasks using a commercial in-ear electroencephalography (EEG) system, the IDUN "Guardian".
Approach: Participants performed versions of two classical workload tasks: an n-back task and a mental arithmetic task. Both in-ear and conventional EEG data were simultaneously collected during these tasks. In an effort to facilitate a more comprehensive comparison, the complexity of the tasks was intentionally increased beyond typical levels. Special emphasis was also placed on understanding the significance of gamma band activity in workload estimations. Therefore, each signal was analyzed across low frequency (1-35 Hz) and high frequency (1-100 Hz) ranges. Additionally, surrogate in-ear EEG measures, derived from the conventional EEG recordings, were extracted and examined.
Main results: Workload estimation using in-ear EEG yielded statistically significant performance levels, surpassing chance levels with 44.1% for four classes and 68.4% for two classes in the n-back task and was better than a naive predictor for the mental arithmetic task. Conventional EEG exhibited significantly higher performance compared to in-ear EEG, achieving 80.3% and 92.9% accuracy for the respective tasks, along with lower error rates than the naive predictor. The developed surrogate measures achieved improved results, reaching accuracies of 57.5% and 85.5%, thus providing insights for enhancing future in-ear systems. Notably, most high frequency range signals outperformed their low frequency counterparts in terms of accuracy validating that high frequency gamma band features can improve workload estimation.
 Significance: The application of EEG-based Brain-Computer Interfaces (BCIs) beyond laboratory settings is often hindered by practical limitations. In-ear EEG systems offer a promising solution to this problem, potentially enabling everyday use. This study evaluates the performance of a commercial in-ear headset and provides guidelines for increased effectiveness.

Optimization of FDM 3D Printer Process Parameters to Minimize Dimensional Errors with PLA Material Using Response Surface Methodology

This paper presents a response surface methodology to fit a second-order response surface model aimed to finding process parameters to minimize length error (LE) and diameter error of the cylindrical shafts (DE) in fused deposition modeling (FDM) using polylactic acid (PLA) material. The process parameters in this study included layer height (LH), which varied from 0.1 to 0.4 mm, and print speed (PS), which ranged from 20 to 60 mm/s, while other parameters were set to constant values. The results showed that optimal process parameters obtained in this study significantly lower dimensional errors than the initial parameterization recommended by UltiMaker Cura software.

Asymptotic Confidence Intervals for MRL and TVaR of the Inverse Weibull Distribution with Type II Censored Data

We consider interval estimation for the parameters, the Mean Residual Life (MRL) and the Tail Value at Risk (TVaR) of the inverse Weibull distribution. We constructed the confidence intervals based on the asymptotic normality of the maximum likelihood estimator in addition to intervals based on the asymptotic Chi-Square distribution of the likelihood ratio (LR) statistic under Type II Censoring. The performance of intervals is investigated and compared using the lower, upper, and total error rates, in addition to expected lengths using simulation. We found that LR intervals showed better overall performance and are more accurate than confidence intervals based on the asymptotic normality of maximum likelihood estimator.

Efficient Flood Detection through Hybrid Machine Learning and Metaheuristic Methods using Sentinel-1

Abstract. Floods are considered among the most destructive natural disasters, requiring precise and timely management. Remote sensing, utilizing diverse satellite imagery data, enables effective monitoring and assessment of flood impacts. In this context, machine learning and deep learning methods, as effective and scalable approaches, can significantly enhance the accuracy of flood detection and management by analyzing remote sensing data, thereby playing a crucial role in mitigating flood-related risks. In this study, to flood detection using Sentinel-1 SAR data, machine learning algorithms, including Random Forest (RF) and Histogram-based Gradient Boosting Decision Tree, were employed, along with two metaheuristic algorithms, Harris Hawks Optimization (HHO) and Ant Colony Optimization (ACO), for hyperparameter optimization. Additionally, to enhance the models' ability to detect flooded pixels and improve overall performance accurately, a pre-trained VGG-16 Neural Network was used as a deep feature extractor. Finally, four ensemble flood detection models—RF-HHO, RF-ACO, HGBDT-HHO, and HGBDT-ACO—were implemented, and their performance was evaluated and compared based on statistical metrics. Based on the obtained results, all four ensemble flood detection models demonstrated excellent performance in the validation and testing phases. The overall accuracy of these models reached over 95% in the validation phase and exceeded 97% in the testing phase. However, the HGBDT-ACO model achieved the highest accuracy and the lowest error rate in detecting flood pixels, making it the best-performing model in this study. Generally, HGBDT models showed a relative advantage over RF models, as they required significantly less time for training while achieving comparable results. Therefore, they were efficient and performed better in terms of computational complexity.

Gramian angular field-based state-of-health estimation of lithium-ion batteries using two-dimensional convolutional neural network and bidirectional long short-term memory

Accurate assessment of the state of health (SOH) of Lithium-ion batteries (LIBs) is crucial for ensuring the stability and reliability of associated equipment. However, in practical applications, current feature extraction techniques face challenges due to process complexity and the difficulty of models in capturing the dynamic evolution of time series data. This study introduces an advanced method for predicting LIBs SOH by integrating two-dimensional (2D) convolutional neural networks (CNN) and bidirectional long short-term memory (BiLSTM) with the Gramian angular field (GAF) technique. Health indicators (HIs) are derived from the incremental capacity and charging voltage curves, which are transformed into two-dimensional images using GAF. Bayesian optimization determines the optimal hyperparameters for the model, which uses image data of two health factors as inputs. The results demonstrate that the proposed dual-channel(CH2) image data input model, combined with voltage data during the charging phase, achieves superior performance, with lower errors and higher accuracy, evidenced by an average root mean square error (RMSE) of 0.0112 and an average mean absolute error (MAE) of 0.0087. The model's generalization capability and comparative analysis with existing methodologies affirm its practical significance.

Revised cross-correlation and time-lag between cosmic ray intensity and solar activity using Chatterjee’s correlation coefficient

This study revisits the cross-correlation between cosmic ray intensity (CRI) and solar activity (SA) by comparing traditional Pearson correlation with Chatterjee’s correlation coefficient. Traditional analyses using Pearson correlation are useful for identifying linear relationships and time lags. However, they may not fully capture more complex interactions in the data. Chatterjee’s correlation coefficient, while sensitive to different types of relationships, including nonlinear ones, provides a complementary perspective on the temporal relationships between CRI and SA. This approach broadens our understanding of potential dependencies, offering additional insights that may not be captured through Pearson correlation alone.The findings reveal that Chatterjee’s correlation complements Pearson’s insights by providing an alternative view of the relationship between cosmic ray intensity (CRI) and solar activity (SA). The results show that Chatterjee’s correlation coefficients are, on average, approximately 45–50% smaller than Pearson’s, which could reflect different sensitivities to the underlying data structure rather than solely indicating a nonlinear component. Additionally, the time lags identified using Chatterjee’s correlation are generally shorter and more consistent across different solar cycles compared to those obtained with Pearson’s correlation, suggesting that CCC may capture temporal patterns in a distinct manner.Further analysis using Dynamic Time Warping (DTW) and Mean Absolute Percentage Error (MAPE) metrics demonstrated that, in more than half of the scenarios considered, alignment based on Chatterjee’s time lags resulted in lower errors and better alignment of the series compared to Pearson’s lags. This indicates that Chatterjee’s method is particularly effective for capturing the immediate and nuanced responses of CRI to SA changes, especially in recent solar cycles.This comprehensive approach provides broader insights into the dynamic interactions between cosmic ray intensity (CRI) and solar activity (SA), highlighting the importance of considering multiple correlation measures, including both linear and nonlinear approaches, in space weather research. The results suggest that Chatterjee’s correlation offers a complementary perspective on these interactions, providing additional details about how SA influences CRI over time, which may not be fully captured by Pearson’s correlation alone.