Initial Data Research Articles

On dynamical system modeling of Learned Primal-Dual with a linear operator $\mathcal{K}$: Stability and convergence properties

Abstract Learned Primal-Dual (LPD) is a deep learning based method for composite optimization problems that is based on unrolling/unfolding the primal-dual hybrid gradient algorithm. While achieving great successes in applications, the mathematical interpretation of LPD as a truncated iterative scheme is not necessarily sufficient to fully understand its properties. In this paper, we study the LPD with a general linear operator. We model the forward propagation of LPD as a system of difference equations and a system of differential equations in discrete- and continuous-time settings (for primal and dual variables/trajectories), which are named discrete-time LPD and continuous-time LPD, respectively. Forward analyses such as stabilities and the convergence of the state variables of the discrete-time LPD to the solution of continuous-time LPD are given. Moreover, we analyze the learning problems with/without regularization terms of both discrete-time and continuous-time LPD from the optimal control viewpoint. We prove convergence results of their optimal solutions with respect to the network state initialization and training data, showing in some sense the topological stability of the learning problems. We also establish convergence from the solution of the discrete-time LPD learning problem to that of the continuous-time LPD learning problem through a piecewise linear extension, under some appropriate assumptions on the space of learnable parameters. This study demonstrates theoretically the robustness of the LPD structure and the associated training process, and can induce some future research and applications.

Leveraging QSP Models for MIPD: A Case Study for Warfarin/INR.

Warfarin dosing remains challenging due to substantial inter-individual variability, which can lead to unsafe or ineffective therapy with standard dosing. Model-informed precision dosing (MIPD) can help individualize warfarin dosing, requiring the selection of a suitable model. For models developed from clinical data, the dependence on the study design and population raises questions about generalizability. Quantitative system pharmacology (QSP) models promise better extrapolation abilities; however, their complexity and lack of validation on clinical data raise questions about applicability in MIPD. We have previously derived a mechanistic warfarin/international normalized ratio (INR) model from a blood coagulation QSP model. In this article, we evaluated the predictive performance of the warfarin/INR model in the context of MIPD using an external dataset with INR data from patients starting warfarin treatment. We assessed the accuracy and precision of model predictions, benchmarked against an empirically based reference model. Additionally, we evaluated covariate contributions and assessed the predictive performance separately in the more challenging outpatient data. The warfarin/INR model performed comparably to the reference model across various measures despite not being calibrated with warfarin initiation data. Including CYP2C9 and/or VKORC1 genotypes as covariates improved the prediction quality of the warfarin/INR model, even after assimilating 4 days of INR data. The outpatient INR exhibited higher unexplained variability, and predictions slightly exceeded observed values, suggesting that model adjustments might be necessary when transitioning from an inpatient to an outpatient setting. Overall, this research underscores the potential of QSP-derived models for MIPD, offering a complementary approach to empirical model development.

Monitoring and analysis of physical activity and health conditions based on smart wearable devices

The rapid growth of the Internet of Things (IoT) brings sweeping changes in various industries. Healthcare industries have become a prime example where the Internet of Healthcare Things (IoHT) is making significant progress, particularly in how we approach real-time patient care. Traditional systems for monitoring older people and people with special needs are frequently expensive, require a large workforce, and fall short of providing real-time data. This paper introduces the “3-Tier Health Care Architecture,” an integrated approach to mitigating these issues. This architecture capitalizes on IoHT technologies and is constructed around three principal tiers: Sensor, Fog, and Cloud. The Sensor Tier employs Health Metrics Acquisition Units (HMAUs) fitted with an nRF5340 Development Kit, capturing an extensive range of health-related metrics via wearable sensors. These metrics are then relayed to the Local Processing Units (LPUs) in Fog Tier, which operates on Raspberry Pi Zero 2 W microprocessors for the initial data processing before forwarding to the cloud. The Cloud Tier uses a hybrid CNN-LSTM Machine Learning (ML) model to perform Real-Time Healthcare Monitoring (RTHM) status assessments and includes an Early Warning System for immediate alert issuance. The proposed architecture is resilient, scalable, and efficient, serving as a fortified and all-encompassing solution for RTHM. This enables quick medical interventions, thus elevating healthcare quality and potentially life-saving.

Track initialisation for multiple formations based on neutrosophic Hough transform

AbstractTrack crossing is a major issue in the initialisation of multiple‐flight trajectories. To solve this problem, the authors propose a track initialisation algorithm. It uses characteristics of formation flight to find centroids of each formation with DBSCAN algorithm. Then, it initialises the track based on the centroid. The author use a Neutrosophic Hough Transform (NHT) method to improve accuracy and computational speed. That helps address errors caused by the approximation of straight lines between true points resulting from the clustering algorithm. The authors made three experiments using track initialisation data from two flight formations with five target aircraft each, over a span of three frames. NHT, Fuzzy HT, Improved Hough Transform (HT) and HT are compared. Results revealed that the average runtime of NHT was 10.2153 s. The F‐measure of NHT was 100.00%, while that of Fuzzy HT was 9.8347 s. The F‐measure of Fuzzy HT was 80.00%. The Improved HT was 12.0723 s. The F‐measure of Improved HT was 11.76% and HT was 13.783 s. And the F‐measure of HT was 6.87%. The authors lost some computation speed to achieve higher prediction accuracy. The accuracy of the NHT is higher than other methods.

Fuzzy expert system design for detecting stunting

Stunting is a chronic nutritional problem that occurs in toddler due to lack of nutritional intake which results in impaired growth toddler. Usually, toddler who experience stunting are characterized by not increasing weight over a long period of time. Application utilization health which makes it easier for users to access information, one of which can be used to identify toddler who are stunted by selecting symptoms. The symptoms experienced by toddlers go through a system known as the system expert. In this research an expert system will be developed that is capable of early detection developmental disorders in toddlers using the Mamdani fuzzy method. The results obtained from this research are an expert system design for early detection of stunting using the Mamdani fuzzy method. The Mamdani fuzzy method was implemented to group the criteria for toddlers who fall into the stunting category or not from the initial data which is still gray because they are still unsure whether to categorize the toddler as having stunting or not. The detection accuracy rate using the Mamdani fuzzy method is 80.87% compared to expert diagnosis.

Вопросы управления траекторией электронного обучения с применением нейросетевых технологий

This article discusses the application of neural networks in managing the educational trajectory in electronic educational systems. A model of learning management in an electronic information and educational environment is built, and an algorithm for managing adaptive learning is described. A test example of a neural network is constructed, showing the capabilities and problems of applying neural networks to the tasks of constructing trajectories of students’ adaptive personalized training. The following stages of constructing a neural network model are implemented: determining the structure of a neural network and its software implementation using the PyTorch deep machine learning library; preparing a test training sample for training the neural network; importing the initial data for training the model using the modules of the Pandas library; training the model using torch.optim Adam, Adamax, and Rprop optimizers; visualizing the obtained data in the form of graphs based on the matplotlib library; exporting the obtained numerical results and analysing the data. The paper also states the problems of preparing a training data set for teaching a neural network. The paper concludes that using neural networks in the field of building personalized adaptive educational trajectories has great prospects and will be one of the urgent tasks in the coming years.

The Association Between Diet Quality and Glycemic Outcomes Among People with Type 1 Diabetes

BackgroundThe amount and type of food consumed impacts the glycemic response and insulin needs of people with type 1 diabetes mellitus (T1DM). Daily variability in consumption, reflected in diet quality, may acutely impact glycemic levels and insulin needs. ObjectiveType 1 Diabetes Exercise Initiative (T1DEXI) data were examined to evaluate the impact of daily diet quality on near-term glycemic control and interaction with exercise. MethodsUsing the Remote Food Photography Method, ≤8 d of dietary intake data were analyzed per participant. Diet quality was quantified with the Healthy Eating Index-2015 (HEI), where a score of 100 indicates the highest-quality diet. Each participant day was classified as low HEI (≤57) or high HEI (>57) based on the mean of nationally reported HEI data. Within participants, the relationship between diet quality and subsequent glycemia measured by continuous glucose monitoring (CGM) and total insulin dose usage was evaluated using a paired t-test and robust regression models. ResultsTwo hundred twenty-three adults (76% female) with mean ± SD age, HbA1c, and body mass index (BMI) of 37 ± 14 y, 6.6% ± 0.7%, and 25.1 ± 3.6 kg/m2, respectively, were included in these analyses. The mean HEI score was 56 across all participant days. On high HEI days (mean, 66 ± 4) compared with low HEI days (mean, 47 ± 5), total time in range (70–180 mg/dL) was greater (77.2% ± 14% compared with 75.7% ± 14%, respectively, P = 0.01), whereas time above 180 mg/dL (19% ± 14% compared with 21% ± 15%, respectively, P = 0.004), mean glucose (143 ± 22 compared with 145 ± 22 mg/dL, respectively, P = 0.02), and total daily insulin dose (0.52 ± 0.18 compared with 0.54 ± 0.18 U/kg/d, respectively, P = 0.009) were lower. The interaction between diet quality and exercise on glycemia was not significant. ConclusionsHigher HEI scores correlated with improved glycemia and lower insulin needs, although the impact of diet quality was modest and smaller than the previously reported impact of exercise.

Application of Ensemble Learning and Feature Selection Models in Energy Expenditure Estimation of Fitness Tracking Devices

The use of fitness tracking devices to monitor energy expenditure (EE) is becoming increasingly popular among consumers. However, the accuracy of EE estimation still needs to be improved due to the lack of accurate data filtering and more efficient data analysis models. To address this issue, we designed an energy expenditure estimation (EEE) model based on smart tracking devices. This paper proposes an algorithm for data reconstruction called PC-DF-RFECV: firstly, Pearson correlation analysis (PC) is used for initial data screening, then Feature Derivation (DF) is applied to reconstruct new features, and finally, the Recursive Feature Elimination algorithm based on cross-validation (RFECV) is employed to select the new features. Furthermore, the EEE model adopts a weighted average (Wei-Voting) strategy to enhance the robustness of the model by integrating the predictions of multiple base learners (XGBoost, LightGBM, Random Forest). During algorithm optimization, a Bayesian hyperparameter optimization technique is employed to fine-tune the model's hyperparameters. Empirical findings indicate that the EEE model, when applied to the dataset collected from motion tracking devices, attains an score of 0.9985, of 1.033, and of 0.865. These results demonstrate superior performance compared to current state-of-the-art research and conventional algorithms including random forests, gradient boosting, and bagging.

Adaptive HD-sEMG decomposition: towards robust real-time decoding of neural drive.

Objective. Neural interfacing via decomposition of high-density surface electromyography (HD-sEMG) should be robust to signal non-stationarities incurred by changes in joint pose and contraction intensity.Approach. We present an adaptive real-time motor unit decoding algorithm and test it on HD-sEMG collected from the extensor carpi radialis brevis during isometric contractions over a range of wrist angles and contraction intensities. The performance of the algorithm was verified using high-confidence benchmark decompositions derived from concurrently recorded intramuscular electromyography.Main results. In trials where contraction conditions between the initialization and testing data differed, the adaptive decoding algorithm maintained significantly higher decoding accuracies when compared to static decoding methods.Significance. Using "gold standard" verification techniques, we demonstrate the limitations of filter re-use decoding methods and show the necessity of parameter adaptation to achieve robust neural decoding.

Boundedness in a chemotaxis-May–Nowak model with exposed state

This paper is concerned with the Neumann initial–boundary value problem of the chemotaxis-May–Nowak model with exposed state for virus dynamics. It is proved that the problem admits a unique global classical solution which is uniformly bounded for all sufficiently smooth initial data in smoothly bounded domains Ω⊂Rn, n≤3.

Predicting the onset of end-stage knee osteoarthritis over two- and five-years using machine learning

ObjectiveIdentifying participants who will progress to advanced stage in knee osteoarthritis (KOA) trials remains a significant challenge. Current tools, relying on total knee replacements (TKR), fall short in reliability due to the extraneous factors influencing TKR decisions. Acknowledging these limitations, our study identifies a critical need for a more robust metric to assess severe KOA. The end-stage KOA (esKOA) measure, which combines symptomatic and radiographic criteria, serves as a solid indicator. To enhance future trials that use esKOA as an endpoint, our study focuses on developing and validating a machine-learning tool to identify individuals likely to develop esKOA within 2 to 5 years. DesignUtilizing the Osteoarthritis Initiative (OAI) data, we trained models on 3,114 participants and validated them with 606 participants for the right knee, and similarly for the left knee, with external validation from the Multicentre Osteoarthritis Study (MOST) involving 1,602 participants. We aimed to predict esKOA onset at 2-to-2.5 years and 4-to-5 years, defining esKOA by severe radiographic KOA with moderate/severe symptoms or mild/moderate radiographic KOA with persistent/intense symptoms. Our analysis considered 51 candidate predictors, including demographics, clinical history, physical examination, and X-ray evaluations. An online tool predicting esKOA progression, based on models with ten and nine predictors for the right and left knees, respectively, was developed. ResultsExternal validation (MOST) for the right knee at 2.5 years yielded an Area Under Curve (AUC) of 0.847 (95 % CI 0.811 to 0.882), and at 5 years, 0.853 (95 % CI 0.823 to 0.881); for the left knee at 2.5 years, AUC was 0.824 (95 % CI 0.782 to 0.857), and at 5 years, 0.807 (95 % CI 0.768 to 0.843). Models with fewer predictors demonstrated comparable performance. The online tool is available at: https://eskoa.shinyapps.io/webapp/. ConclusionOur study unveils a robust, externally validated machine learning tool proficient in predicting the onset of esKOA over the next 2 to 5 years. Our tool can lead to more efficient KOA trials.

Diseño de un sistema de agitación para un biorreactor de 1m3 de capacidad para la producción de ácido glicérico a partir de glicerol por fermentación microbiana de Acetobacter tropicalis

This work presents the design of an agitation system for a 1m3 capacity bioreactor intended to produce glyceric acid (GA) through microbial fermentation of A. tropicalis using 20% v/v glycerol as the primary substrate. Building on previous studies that evaluated ideal conditions for glyceric acid production, initial data were utilized to identify the impeller type and perform subsequent calculations, including impeller Reynolds number, power consumption, and mixing time. Once the agitator type was specified, standard geometric relationships were employed to size components of the agitation system, such as tank size, baffle plates, and agitator dimensions. Power consumption calculations were compared with results from other studies on fermentation media with similar rheological characteristics. The agitator's power consumption for mixing 1m3 of medium was determined to be 8.35 kW, achieving 95% homogenization in 5.14 seconds, with a mixing time error of approximately 10%, operating at an agitation speed of 500 rpm. Keywords: agitation system; glyceryc acid; bioreactor

Streamlining Telecommunications Center and Interfacility Patient Throughput to a Pediatric Emergency Department by Utilizing an Electronic Handoff: A Quality Improvement Initiative.

Effective handoffs are critical for patient safety and high-quality care. The pediatric emergency department serves as the initial reception for patients where optimal communication is crucial. The complexities of interfacility handoffs can result in information loss due to lack of standardization. The aim of our project was a 50% reduction in monthly calls routed through the communication center from 157 to 78, for interfacility transfers to the emergency department from outpatient sites within our organization over a 1-year period, through utilization of an electronic handoff activity. We designed a quality improvement project in a tertiary care pediatric hospital to improve the process of interfacility transfer. The initiative aimed to streamline the transfer of patients from ambulatory, urgent care, and nurse triage encounters to the pediatric emergency department by using the electronic health record. The primary outcome measure was number of monthly calls received by the telecommunications center for these transfers.Our process measure was tracked by measuring the utilization of the electronic handoff. In addition, the number of safety events reported because of information lost through using the electronic handoff served as a balancing measure. One year after the enterprise-wide rollout of the handoff, the telecommunications center was receiving an average of 29 calls per month versus 157 at time of study initiation, a decrease of 81.5%. Monthly usage increased from zero to an average of 544 during the same period. The project was continued after the initial 12-month data collection and demonstrated stability. Our initiative facilitated the safe and efficient transfer of patients and streamlined workflows without sacrificing quality of patient care. Our telecommunications center has been freed up for other tasks with fewer interruptions during patient throughput. Next steps will analyze the encounters of transferred patients to further optimize patient flow at our organization.

Panaxynol improves crypt and mucosal architecture, suppresses colitis-enriched microbes, and alters the immune response to mitigate colitis.

Ulcerative colitis (UC) is an idiopathic inflammatory disease of the large intestine, which impacts millions worldwide. Current interventions aimed at treating UC symptoms can have off-target effects, invoking the need for alternatives that may provide similar benefits with less unintended consequences. This study builds on our initial data, which showed that panaxynol-a novel, potent, bioavailable compound found in American ginseng-can suppress disease severity in murine colitis. Here we explore the underlying mechanisms by which panaxynol improves both chronic and acute murine colitis. Fourteen-week-old C57BL/6 female mice were either given three rounds of dextran sulfate sodium (DSS) in drinking water to induce chronic colitis or one round to induce acute colitis. Vehicle or panaxynol (2.5 mg/kg) was administered via oral gavage three times per week for the study duration. Consistent with our previous findings, panaxynol significantly (P < 0.05) improved the disease activity index and endoscopic scores in both models. Using the acute model to examine potential mechanisms, we show that panaxynol significantly (P < 0.05) reduced DSS-induced crypt distortion, goblet cell loss, and mucus loss in the colon. 16S Sequencing revealed panaxynol altered microbial composition to suppress colitis-enriched genera (i.e., Enterococcus, Eubacterium, and Ruminococcus). In addition, panaxynol significantly (P < 0.05) suppressed macrophages and induced regulatory T-cells in the colonic lamina propria. The beneficial effects of panaxynol on mucosal and crypt architecture, combined with its microbial and immune-mediated effects, provide insight into the mechanisms by which panaxynol suppresses murine colitis. Overall, this data is promising for the use of panaxynol to improve colitis in the clinic.NEW & NOTEWORTHY In the current study, we report that panaxynol ameliorates chemically induced murine colitis by improving colonic crypt and mucosal architecture, suppressing colitis-enriched microbes, reducing macrophages, and promoting the differentiation of regulatory T-cells in the colonic lamina propria. This study suggests that this novel natural compound may serve as asafe and effective treatment option for colitis patients.

Sentiment Analysis on the Quality of Public Services with User Satisfaction Prediction of YuhSinau Application Managed by BKPSDM Kabupaten Kebumen Using LSTM Method

The quality of public services is critical in providing effective and responsive governance in an increasingly digital society. The development of the YuhSinau application by the Personnel and Human Resource Development Agency (BKPSDM) of Kebumen Regency offers an innovative response to the growing need for e-learning solutions for local government civil servants (Pegawai Negeri Sipil or PNS). However, determining the app’s effectiveness and user satisfaction is critical. This demands a thorough sentiment analysis in order to acquire insights into users’ thoughts and opinions about the quality of public services supplied by YuhSinau. The Long Short-Term Memory (LSTM) approach is used in this article to examine feelings and forecast customer pleasure. Data collection from multiple sources, initial data preprocessing, LSTM model construction, training, validation, and prediction are all part of the process. The results show that the model has some drawbacks, most notably its failure to appropriately explain variation in target data due to a negative R-squared value. Enhancements to the LSTM architecture, hyperparameter adjustment, and the use of more diverse and representative training data are proposed to improve the model. Continuous review and responsiveness to user comments are critical for improving the quality of public services via the YuhSinau application.&#x0D; Keywords: sentiment analysis, public service quality, user satisfaction, YuhSinau application,LSTM method

Alkaloid chemistry in pill-millipedes: Defensive secretion in two species of Typhloglomeris Verhoeff, 1898 (Diplopoda, Glomerida, Glomeridellidae)

AbstractMillipedes of the order Glomerida are known to produce quinazolinone alkaloids as defensive substances. However, chemical data are rather scarce and all hitherto available publications refer to a single glomeridan family, the Glomeridae. To contribute to the knowledge of the chemical ecology of glomeridans, we collected species, Typhloglomeris coeca Verhoeff, 1898 and T. varunae Makarov, Lučić, Tomić &amp; Karaman, 2003, from a second family – Glomeridellidae. The defensive exudates were extracted in methylene chloride and analyzed by gas chromatography-mass spectrometry. Our results show that the secretions of the two glomeridellids contain the quinazolinone alkaloid homoglomerin (2-ethyl-1-methylquinazolin-4(1H)-one) as the sole defensive constituent. This report provides initial data on the chemoprofiles of members of the family Glomeridellidae and supports the phenomenon of chemical homogeneity of glomeridan defensive secretions. So far, all analyzed representatives of the order possess glomerin and/or homoglomerin as defensive equipment. The phylogenetic origin of alkaloidal compounds in the secretions of millipedes in general, and particularly the origin of quinazolinone alkaloids in Glomerida is discussed. Considering that the quinazolinone alkaloid defensive secretions may be the same (or very similar) in other Glomerida, our results have important implication for further studies of defensive secretions in these arthropods. Also, as data are available only for a limited number of glomeridan and colobognathan species, additional analyses are necessary to elucidate the evolution of alkaloid defensive secretion in Diplopoda.

On radially symmetric oscillations of a collisional cold plasma

We study the influence of the friction term on the radially symmetric solutions of the repulsive Euler–Poisson equations with a non‐zero background, corresponding to cold plasma oscillations in multiple spatial dimensions. It is shown that for any arbitrarily small non‐negative constant friction coefficient, there exists a neighborhood of the zero equilibrium in the norm such that the solution of the Cauchy problem with initial data belonging to this neighborhood remains globally smooth in time. Moreover, this solution converges to zero as . This result contrasts with the situation of zero friction, where any small deviation from the zero equilibrium generally leads to a blow‐up. Our method allows us to estimate the lifetime of smooth solutions. Further, we prove that for any initial data, one can find such a coefficient of friction that the respective solution to the Cauchy problem keeps smoothness for all and converges to zero. We also present the results of numerical experiments for physically reasonable situations, which allows us to estimate the value of the friction coefficient, which makes it possible to suppress the formation of singularities of solutions.

A semi-quantitative histochemical method for assessment of biochemical responses to osmotic stress in Coffea arabica leaf disks.

A simple method set for assessing biochemical changes associated with osmotic stress responses was developed using coffee (Coffea arabica L.) leaf disks. Stress was induced by polyethylene glycol (PEG) exposure. Quantitative evaluation of tissue physiological stress parameters was carried out using analytical methods to validate the conversion of classic qualitative histochemical tests for localizing lipid peroxidation, hydrogen peroxide, and total xanthine alkaloids into semi-quantitative assays. Relative electrolyte leakage (EL%) and chlorophyll content (SPAD index) were also recorded. EL% levels of treated disks were higher than those of control ones, whereas SPAD indexes were comparable. Histochemical localization indicated that levels of lipid peroxidation, H2O2, and total xanthines were also higher under osmotic stress than in control conditions. Semi-quantitative data obtained by image processing of histochemical staining consistently matched quantitative evaluations. Chromatographic analyses revealed that theophylline and caffeine concentrations increased in the presence of PEG, whereas theobromine remained constant in relation to the control. The methods herein described can be useful to rapidly acquire initial data regarding biochemical osmotic stress responses in coffee tissues based on simple staining and imaging steps. Moreover, it is likely that the same method may be applicable to other types of stresses and plant species upon minor adjustments.

Beyond ethics: How perceived organizational support amplifies the impact of ethical HRM on employee commitment and performance

This study explores the dynamic relationship between ethical human resources management (HRM) strategies, the level of commitment an employee feels towards their organization, and their job performance, paying particular attention to how employees’ perceptions of the support they receive from their organization can influence these interactions, especially during challenging times. Drawing on a sample of full-time non-executive Indonesian employees, the research employs descriptive statistics for initial data analysis, followed by structural equation modeling (SEM) to test the proposed hypotheses rigorously. The investigation reveals a positive relationship between ethical HRM and employee performance (EP) and organizational commitment (OC). Additionally, OC emerges as a pivotal mediator in the ethical HRM-EP link. Notably, employees’ organizational support perception (EOSP), often assumed to enhance positive organizational outcomes, displays a surprising negative moderating effect when combined with OC, suggesting a more intricate relationship than traditionally posited. These findings enhance our comprehension of how ethical HRM practices function in times of crisis, questioning conventional beliefs regarding the influence of organizational support. The study’s methodological approach, combining descriptive and advanced statistical analyses, provides a robust framework for understanding these complex relationships. This research holds significant implications for HRM practices, particularly in crisis response and management, indicating a need for nuanced support strategies that reflect the complexity of employee-organization dynamics.

Detect with Style: A Contrastive Learning Framework for Detecting Computer-Generated Images

The detection of computer-generated (CG) multimedia content has become of utmost importance due to the advances in digital image processing and computer graphics. Realistic CG images could be used for fraudulent purposes due to the deceiving recognition capabilities of human eyes. So, there is a need to deploy algorithmic tools for distinguishing CG images from natural ones within multimedia forensics. Here, an end-to-end framework is proposed to tackle the problem of distinguishing CG images from natural ones by utilizing supervised contrastive learning and arbitrary style transfer by means of a two-stage deep neural network architecture. This architecture enables discrimination by leveraging per-class embeddings and generating multiple training samples to increase model capacity without the need for a vast amount of initial data. Stochastic weight averaging (SWA) is also employed to improve the generalization and stability of the proposed framework. Extensive experiments are conducted to investigate the impact of various noise conditions on the classification accuracy and the proposed framework’s generalization ability. The conducted experiments demonstrate superior performance over the existing state-of-the-art methodologies on the public DSTok, Rahmouni, and LSCGB benchmark datasets. Hypothesis testing asserts that the improvements in detection accuracy are statistically significant.