Intelligent Decision Technologies

As the secondary ship market becomes increasingly active, achieving precise ship valuation while fully accounting for market fluctuations has grown increasingly important. This paper establishes a price assessment framework integrating static valuation with dynamic market adjustments: the static component employs an ACO–FA-optimized BP neural network to derive benchmark prices based on individual vessel characteristics; the dynamic component constructs a price index combined with oil prices and freight rates into a multivariate time series. A GRU model captures market adjustment factors, which are then fused with static valuations via Kalman filtering to generate final transaction prices. Empirical results based on Chongqing's 2020–2025 dry bulk carrier transaction data demonstrate that this model significantly enhances second-hand vessel price assessment while simultaneously delivering static valuations, market adjustments, and composite prices. It provides buyers and sellers with a well-explained quantitative basis for pricing, negotiation, and investment decisions across varying market conditions.

The Industrial Internet of Things (IIoT) is characterized by the instantaneous communication of diverse critical data across complex, resource-limited networks. To maximize the efficiency and accuracy of task scheduling in such networks, this study introduces a novel task scheduling framework based on the Adaptive Lotus Effect Optimization Algorithm (ALEOA). ALEOA perfectly combines the self-cleaning mechanism and learning function of the Lotus effect with the global search and learning capabilities of the Particle Swarm Optimization (PSO) algorithm to establish equilibrium between local and global search. ALEOA can adaptively adjust routing and task scheduling to support critical data communication and a dynamic network. Results from experiments on typical benchmarking equations and simulations of the IIoT communication environment show that the proposed ALEOA-based model yields higher task acceptance rates, lower communication delays, and improved resource utilization than traditional metaheuristic methods. Statistical analyses using the Wilcoxon Signed Rank test confirmed the effectiveness of this enhancement.

Imbalanced classification remains a critical challenge in decision-sensitive domains such as healthcare, finance, and cybersecurity, where minority class recognition is often paramount. This paper introduces FRSTU-Forest, a novel hybrid framework that integrates K-Nearest Neighbor (k-NN) imputation, Fixed Random State Undersampling (FRSTU), and Random Forest to enhance both minority class detection and model reproducibility. Unlike conventional undersampling, FRSTU applies deterministic sampling with a fixed random seed, ensuring consistent training subsets across runs and significantly reducing performance variance. The framework was comprehensively evaluated on seven benchmark datasets with moderate imbalance ratios (1.25–3.36) and rigorously tested on synthetic datasets with extreme imbalance ratios up to 1:100, high dimensionality (100 features), and substantial label noise (20%). FRSTU-Forest consistently outperformed baseline models (RF, k-NNimp+RF, RSTU+RF), achieving an average accuracy of 87.88%, minority-class F1-score up to 99.78%, and Cohen’s Kappa of 0.86 on benchmark datasets. More importantly, under extreme imbalance conditions (1:100 ratio), it maintained a balanced accuracy of 0.807 with 100 features, demonstrating remarkable robustness. Statistical significance was confirmed via the Bonferroni-Dunn test ( p = 0.0015 ), while computational efficiency remained practical with an average runtime of 2.01 seconds per dataset. In real-world applications, the model achieved 96.92% recall on breast cancer detection and substantially improved credit risk classification. These results affirm that FRSTU-Forest provides a reliable, reproducible, and robust decision-support tool for imbalanced data environments, particularly effective in scenarios with severe class imbalance, high dimensionality, and noisy labels.

Liquefied Petroleum Gas (LPG) is widely used in households, commercial establishments industrial sectors. LPG is highly flammable and poses a risk of explosion if not handled properly. Hence, an early gas leakage detection system is crucial for ensuring safety in various environments. In this article, a Deep Learning (DL)-based LPG Gas leakage detection and Alert system named Quasi-Recurrent Neural Network with Addax Wolf Bird Optimization Algorithm (QRNN_AWBOA) is proposed using the gas leakage data. In this approach, the median normalization method is utilized to normalize the raw data. Then, a fusion model named Deep Neural Network (DNN) with Neyman similarity is utilized for the feature fusion process. Then, the data is augmented using oversampling. Later, the detection process is carried out using the Quasi-Recurrent Neural Network (QRNN) model. The QRNN effectively trained the Addax Wolf Bird Optimization Algorithm (AWBOA). Finally, an alert is sent to the NG112 authorities if the leakage is present. This detection system attained the lowest Mean Squared Error (MSE) of 0.016, Mean Absolute Percentage Error (MAPE) of 0.056, Root Mean Squared Error (RMSE) of 0.128, and Weighted Absolute Percentage Error (WAPE) of 0.057.

This framework is a hybrid combination of AI and blockchain for dance performance assessment. The model it uses, the hybrid CNN/LSTM model, evaluates posture accuracy and timing for the posture, while style recognition requires a Transformer-based encoder to learn long-range dependencies. The Expressiveness Index measures the expressiveness and dynamics shown by a performer through a regression head. A blockchain ensures that the performance results can secure and verifiable credentials without forging them. Real-time feedback helps dancers to improve faster and more effectively. The model is trained on a variety of dance styles to ensure that it is flexible and applicable to different forms of dance training. The evaluation achieved high-performance metrics, showing the strength of the system for objective performance measurements, accuracy of 95.52%, Precision 94.56%, Recall 94.87%, and F1-Score of 94.71%. In the exciting angle of enhancing the integration of AI and blockchain into the performing arts and their visual presentation, TransCNN + DSSS signifies a secure, transparent, and scalable means of providing dance education.

The distributed, heterogeneous, and shared security risks of power equipment data across its entire lifecycle limit the facilitation and integrated sharing of data across the entire lifecycle. This paper proposes a machine-learning-based secure data-sharing model for power equipment data during its lifecycle. The development of the proposed model includes multi-source data merging from the operation, inspection, and maintenance process into one data format through semantic mapping, in the unified structure of the data so that machine-learning is used for feature extraction and risk-prediction for dynamic access control and the adaptive encryption and de-sensitization balance between risk mitigation and data sharing is still maintained. The full-process monitoring and feedback monitoring to detect anomalous behavior can also optimize the polices in real-time. The experimentation provides an assigned data penetration rate of 96.3% for the disconnector. The leakage rate of sensitive information was reduced to 1.8% once the risk level was increased to extremely high. This separation alleviates the conflict of data security and data sharing by providing original research efforts for intelligent information and database systems.

Major health emergencies often trigger rapid changes in public risk perception, amplified by the widespread dissemination of information through intelligent networks. Accurately capturing these dynamics and assessing the authenticity of online information is crucial for effective crisis management. This research proposes a data-driven framework to analyze the dynamic evolution of public risk perception and the reliability of information dissemination during viral disease emergencies. The approach integrates association rule mining with an Artificial Lizard Search-driven Attention-refined Recurrent Neural Networks (ALS-Att-RNN) model to identify the public perception key risk factors and key risk chains of different risk levels, as well as to evaluate the public opinion situation. To ensure reliable model performance, data preprocessing was carried out using Z-score normalization, which standardized the input data and enhanced the accuracy of subsequent analysis. An association rule mining algorithm is designed to handle non-Boolean and continuous data, enabling the extraction of significant correlations among misinformation signals, key viral risk indicators, and public perception trends. The framework identifies critical misinformation chains, trustworthy information sources, and key perception risk factors, providing a quantitative evaluation of information authenticity across social and intelligent network platforms. Empirical results show that the proposed method improves assessment accuracy (0.98) compared to conventional models. This framework provides a computationally efficient and interpretable tool for health authorities to identify misinformation chains, key perception risk factors, and public response patterns, supporting timely interventions and guiding informed decision-making to mitigate viral disease risks.

This research work introduces a hybrid AI model comprising XGBoost for static prediction, LSTM networks for temporal forecasting, and SHAP for interpretable explainability. One of the major public health issues is retention in childhood malnutrition, especially in preschool children, which greatly impacts their growth, cognitive development and health conditions later in life. The conventional methods, based on descriptive statistical analysis and basic machine learning approaches, mainly utilise static predictors to classify malnutrition cases. Such practices tend to overlook the changes over time and do not offer any explanation, leading to their limited applications in preventive measure interactions. The model not only considers the current nutritional status but also predicts future trends by identifying important demographic and dietary components that underlie the predictions. The AI framework suggested produced 98.86% accuracy, AUC = 0.991, and RMSE = 0.082 when validated on CHNS data, which was significantly greater than the performance of Logistic Model Trees (91–95%) and Random Forests (94–98.6%), resulting in up to a 4.1% improvement. It thus confirmed the model's outstanding predictive capability. In contrast to previous works that were merely concerned with malnutrition type or anaemia classification, the present approach guarantees both accuracy and usability by integrating temporal prediction with Explainable AI. Consequently, the integration of predictive modelling with practical policy implications enables the early identification of children at risk and the implementation of targeted interventions, thereby providing a scalable and evidence-based contribution to childhood nutrition research. The combined model achieved an accuracy of 98.86%, an F1 score of 0.988, and an AUC of 0.991, which are extremely high values for predicting and highlighting the practicality of early detection and intervention in childhood malnutrition.

This paper presents a novel prediction model for Regional Economic Development (RED) levels using the Self-Organizing Map (SOM) algorithm. It posits that the SOM algorithm can effectively forecast RED levels by processing multidimensional datasets that include economic, social, and environmental indicators. The approach involves constructing an Evaluation Index System (EIS) comprising 14 indicators across three domains: financial performance, social conditions, and ecological sustainability. The performance of the SOM model is compared to the traditional Support Vector Regression (SVR) model. The analytical tools utilized in this study include the SOM neural network and the SVR model, both applied to predict the Gross Domestic Product (GDP) and the Consumer Price Index (CPI) in Region C. The SOM model achieves a relative error of 0.01% in predicting GDP and 0.12% in predicting CPI, outperforming the SVR model. Additionally, the model is applied to forecast the economic development levels of two Chinese provinces from 2026 to 2035, revealing significant regional disparities. The findings suggest that the SOM algorithm is a promising tool for predicting RED, providing valuable insights for policy-making processes.

The region encircled by the circle of radius u around an interval-valued intuitionistic fuzzy set is known as the circular interval-valued intuitionistic fuzzy set. Since intervals are used to represent them, interval-valued intuitionistic fuzzy sets can deal with ambiguity and uncertainty in a variety of ways, however, the area encircling them is not taken into account. As the area around it is taken into consideration, a circular interval-valued intuitionistic fuzzy set is therefore more able to handle uncertainty than an interval-valued intuitionistic fuzzy set. Effective decision-making can result in progress, happiness, and prosperity, making it crucial for both professional and personal success. All of the fuzzy extension's distance measures have applications in a variety of fields, including pattern recognition, clustering, medical diagnosis, and more. As a result, distance measures for circular interval-valued intuitionistic fuzzy sets are presented, and the proposed distance measures are used to solve a group decision-making challenge that is selected from the literature. TOPSIS is the method used for tackling the decision-making problem. The same problem is assessed using the existing distance measures in order to further validate the suggested distance measures of circular interval-valued intuitionistic fuzzy sets, and the outcomes are contrasted. In addition, to verify the robustness of the proposed distance measures on circular interval-valued intuitionistic fuzzy environment, sensitivity analysis is done using the operators ⊗ m i n and ⊗ m a x .

Intelligent Tutoring Systems face significant challenges in dynamically adapting to diverse student cognitive states (e.g., knowledge gaps, learning pace, and engagement levels) while maintaining high confidence in pedagogical decisions. Traditional rule-based or static Machine Learning (ML) models often fail to generalize across different learners and subjects. A confidence-aware Meta-Reinforcement Learning (Meta-RL) framework is proposed, allowing for fast adaptation to individual student needs with quantifiable uncertainty estimation. The Proposed Method (PM) leverages meta-learning to pre-train a policy on a distribution of simulated and real-world student interactions, allowing rapid fine-tuning with minimal data for new learners. The framework incorporates Bayesian Neural Networks (BNN) to assess prediction confidence, ensuring that tutoring actions (e.g., hint provision or problem difficulty adjustment) are personalized and reliable. Experiments on large-scale educational datasets (e.g., ASSISTments, MOOC logs) demonstrate that the model outperforms baseline methods (e.g., deep RL, non-adaptive ITS) in learning gain (+12%) and student engagement (+18%), with real-time deployment feasibility on edge devices. This work bridges the gap between high-speed adaptation and high-confidence AI in education, offering a scalable solution for next-generation ITS.

Job scheduling has become one of the most challenging issues in the research field of data centers within the cloud computing sector. Green cloud computing is a paradigm that employs efficient techniques to significantly reduce carbon emissions, CPU frequency scaling, and overall energy consumption. The large volume of data processed necessitates the consideration of data center strategies to mitigate energy consumption, along with other factors such as environmental impact, operational cost, and system reliability. Thus, this research aims to develop a new job scheduling scheme in a big data-assisted green cloud environment by formulating a multi-objective optimization problem that considers makespan, power consumption, latency, and resource utilization. For handling big data, the MapReduce framework is employed. To implement the green cloud, a Hybrid Firefly Water Strider Optimization (HFWSO) algorithm is developed by combining the Firefly Algorithm (FFA) and Water Strider Algorithm (WSA). This algorithm optimizes the number of servers and jobs allocated based on data volume, ensuring efficient job distribution across servers. Task scheduling is performed to solve a multi-objective function under various constraints. The makespan of the developed model is 148.3 s, energy consumption is 6.72 kWh, and resource utilization is 76.50%. Thus, the findings demonstrate that it effectively minimizes the energy consumption in data centers and also appropriately allocates jobs to machines. These comparative findings also reinforce the suitability of HFWSO as a powerful metaheuristic for multi-objective optimization in energy-efficient cloud computing environments, making it a promising candidate for practical applications requiring balanced trade-offs among makespan, energy consumption, latency, and resource utilization.

Accurate and timely classification of white blood cells (WBCs) is crucial for diagnosing a myriad of hematological disorders, including leukemia. While deep learning models, particularly Convolutional Neural Networks (CNNs), have shown promise in automating this task from microscopic blood smear images, their performance can be hindered by complex backgrounds and intra-class variations. This paper proposes a novel segmentation-enhanced classification framework that synergistically combines classical image processing with deep learning. Our approach first employs a fixed-parameter Canny edge detection and contour-based algorithm to segment the WBC foreground. Subsequently, a learnable blending layer intelligently fuses the segmented foreground with the original image, allowing the downstream CNN to leverage both focused object information and contextual cues. We meticulously document our experimental journey, including initial attempts to train Canny parameters which proved unstable. The proposed model, featuring fixed segmentation and a learnable blending factor ( α ), was evaluated on a public blood cell image dataset for cancer detection. It achieved a test accuracy of 99.93%, outperforming a baseline CNN (99.07%). The blending factor α converged to approximately 0.63, indicating an optimal balance between the segmented and original image content. Furthermore, the segmentation-enhanced model demonstrated faster inference times. This work underscores the potential of hybrid approaches, particularly the utility of a learnable blending mechanism to effectively integrate classical segmentation outputs into neural networks for improved classification.

In the era of smart technology, the integration of electronics with textiles has given rise to intelligent clothing garments embedded with electronic systems capable of sensing, processing, and responding to environmental or physiological data. However, ensuring both sensor accuracy and user comfort remains a significant challenge. To address this, the present research aims to design and implement an embedded smart clothing system that combines wearable sensor technology with soft-textile materials to enable real-time monitoring of body temperature while maintaining high wearing comfort. The proposed system utilises strategically placed, skin-loose temperature sensors embedded into infant-friendly fabrics, minimising irritation and enhancing mobility by eliminating the discomfort commonly associated with skin-tight sensors. Custom-designed smart garments equipped with these sensors were used for data collection, followed by pre-processing techniques such as normalisation and one-hot encoding. At the core of the system lies a low-power embedded microcontroller that wirelessly collects, processes, and transmits sensor data, ensuring energy-efficient operation suitable for everyday infant wear. To address the reduced precision of individual loose sensors, a deep learning-based temperature estimation model was developed. This model aggregates multichannel sensor inputs and processes using a Komodo Mlipir fine-tuned Stochastic Temporal Convolutional Network (KM-STCN), enabling real-time estimation and correction of body temperature readings. Implemented using Python, the experimental results demonstrate that the combined use of multiple skin-loose sensors and the deep learning model achieves high temperature estimation accuracy, recall, F1-score, and precision rates exceeding 93%, while significantly enhancing garment comfort. This research highlights the potential of integrating embedded systems, infant garment design, and AI-based signal processing to develop intelligent clothing that effectively balances health-monitoring functionality with everyday wearability in healthcare technologies.

This article investigates decentralized blockchain technology to improve the management of sustainability in prefabricated buildings. By utilizing a public blockchain and a private blockchain together, both real-time recording and management of construction-related data, and contract management-applied automation are accomplished. Smart contracts, Merkle trees, and Proof of Work (PoW) consensus mechanisms guarantee the security, reliability, and transparency of data, especially with respect to those categories including material qualifications, construction progress, and quality inspections. The Hyperledger Fabric platform incorporates encryption technology combined with smart contracts for additionally protecting the privacy of financial data, construction progress, and supply chain data through use of the AES-256 algorithm encryption and strict data audit monitoring controls. Moreover, decentralized blockchain technology improves transparency and controllability of data systems, effectively decreasing the risk of data compromise, and optimizing the management of the supply chain through smart contracts, hash value technology, and other means. Tracking time through the average supply chains is 6.1 s, the average accuracy for fulfillment of contracts is 98.36%, the average success rate for smart contract execution is 98.84%, and the average timeliness of information flow is 95.12%. This article promotes the digital transformation and sustainable development of the construction industry by improving the prefabricated construction sector's capacity for sustainable development and offering fresh concepts for data management and monitoring.

The Industrial Internet of Things (IIoT) facilitates the large-scale interconnection of heterogeneous devices and processes. Nevertheless, scheduling hundreds of latency-sensitive, resource-intensive tasks across distributed systems efficiently remains a challenge. State-of-the-art swarm intelligence and evolutionary approaches often suffer from premature convergence, high computational costs, and unreliable output in dynamic network environments. In this study, we propose a Multi-Strategy Ivy Algorithm (MS-IA) for low-latency and reliable task scheduling in IIoT environments. MS-IA adopts the adaptive growth and propagation behavior of ivy plants, which consists of four strategic extensions: adaptive perturbation, adaptable growth velocity, the fish-aggregation device concept, and hybridization with differential evolution. By incorporating these mechanisms into the IIoT task scheduling model, MS-IA maintains a proper exploration-exploitation balance, thereby reducing makespan and power consumption while maximizing system reliability and throughput. Through comprehensive testing of multi-scale IIoT workloads, it has been certified that MS-IA demonstrably outperforms state-of-the-art scheduling algorithms.

Traditional rule-based systems struggle with complex power grid faults due to limited flexibility and dynamic response. This study proposes an automated fault disposal approach using a power grid fault knowledge graph combined with reasoning algorithms. The system models equipment, fault types, and disposal steps as a graph, where nodes represent entities and edges denote fault relations. Real-time monitoring data feeds into the graph, enabling automated reasoning to identify fault causes, affected areas, and disposal options. Based on this, the system dynamically generates accurate disposal plans with high automation, reducing manual intervention, and supports over 80% of fault types with a maximum response time of 3.5 s, significantly improving diagnosis accuracy and emergency response efficiency. This method offers a robust technical path for intelligent grid operations.

Small and medium-sized enterprises (SMEs) face escalating cyber threats and often lack the technical and financial capacity to deploy resource-intensive security solutions. This study evaluates the effectiveness of Artificial Intelligence (AI) and Machine Learning (ML) models for lightweight intrusion detection systems (IDS) tailored to SME environments. Six supervised algorithms including Logistic Regression, Naïve Bayes, Random Forest, Multi-Layer Perceptron (MLP), LightGBM, and XGBoost, were trained and tested on the TON_IoT dataset. Performance was assessed using standard metrics (accuracy, precision, recall, F1-score, ROC-AUC) alongside computational efficiency indicators (training time, inference latency, and model size). Results show that tree-based ensemble models, particularly XGBoost (accuracy = 99.07%, ROC-AUC = 0.9996) and LightGBM (accuracy = 99.00%, ROC-AUC = 0.9993), achieved superior detection performance while maintaining minimal computational cost. In contrast, Logistic Regression and Naïve Bayes exhibited faster training but lower detection accuracy (<76%). These findings confirm that AI-driven ensemble models can deliver both accuracy and efficiency, making them ideal for deployment in resource-constrained SMEs. Limitations of this study include reliance on a single dataset and supervised frameworks. Future work will extend to cross-dataset validation and real-world SME applications to enhance robustness and generalizability.

With the increasing demand for laboratory safety management, precise detection and graded early warning of misconduct become more critical. Traditional manual monitoring methods suffer from low efficiency, limited coverage, and difficulty handling complex scenarios. To address this, this paper proposes a laboratory misconduct graded early warning model based on object detection algorithms and Graph Convolutional Networks. The model uses You Only Look Once version 9, combined with Convolutional Block Attention Mechanisms, to enhance key feature extraction and accurately identify misconduct. Meanwhile, the Graph Convolutional Network explores spatial correlations between behaviors, and gated recurrent units capture temporal dynamic features to implement graded risk warning. The experimental evaluation showed a minimum loss of 0.027 after 120 iterations, demonstrating superior performance compared with the comparison models, which recorded loss values of 0.24, 0.25, and 0.32. In graded early warning tests, the model reaches an accuracy of 95.62%, with precision and recall exceeding 92%, clearly higher than the highest values of comparison models at 88.21% and 88.01%. These results indicate that the model can achieve precise detection and graded early warning of laboratory misconduct, providing an intelligent solution for laboratory safety management and promoting efficient and accurate safety monitoring.

In response to the challenges of multimodal data analysis in disaster events, this study proposes a two-stage technical framework of “feature alignment evidence fusion”. A cross-modal contrastive learning framework (PMCL) utilizing agents is constructed during the feature alignment stage, which achieves cross-modal feature collaboration through a bimodal Transformer encoder, agent sample collaborative optimization, and geometric constraints. In the fusion decision-making stage, a Cross-modal Enhanced Fusion Network (CEFN) is constructed, and Dirichlet distribution parameterization, projection distance evaluation, and an adaptive fusion mechanism are used to address semantic uncertainty. Experiments have shown that PMCL achieved an accuracy rate of 85% on crisis multimodal information classification datasets, which was 30% higher than the single modal baseline. The accuracy of CEFN in this dataset task was 1.16% higher than that of suboptimal models, and the conflict loss function still controlled performance degradation within 3.34% under 100% inconsistent samples. In addition, PMCL's multimodal pre-training initialization strategy improved the accuracy of the model by 7.1%. This study provides an efficient and interpretable technical solution for disaster emergency response, which has important practical significance for multimodal data-driven intelligent disaster reduction decision-making.

The ever-growing number of energy requirements needed in cloud data centers has led to the rising use of intelligent, sustainable and environmental scheduling algorithms. The study proposes a new hybrid algorithm that is a fusion of Proximal Policy Optimization (PPO) and Round Robin (RR), and is called PPO-RR, to enhance the efficiency of virtual machine (VM) scheduling and reduce the amount of energy consumption in a virtualized cloud system. The suggested approach follows the strategy of reinforcement learning: using the feedback about the system state, it can actively optimize the scheduling decisions to improve resource utilization and eliminate unnecessary waste of power in idle conditions.Experiments were carried using an artificially created cloud environment dataset. The PPO-RR model was seen to have shown significant performance gains over the conventional RR and heuristic scheduling techniques. To be more precise, PPO-RR achieved the accuracy of 95.3%, precision of 94.6%, recall of 95.4, and F1-score of 94.9. Such findings demonstrate the superiority of the model in its capacity to assign the VMs to the hosts appropriately during different workloads.With regard to energy measures, PPO-RR saved 1520 kWh to 1235 kWh translating to about 56.3 percent of energy saved as compared to conventions practices. The PPO-RR confusion matrix shows that the True Positives (TP) are 420, True Negatives (TN) are 360, False Positives (FP) are 15 and False Negatives (FN) are 25 which means that the accuracy of the classification of the decisions in scheduling is high.The model offers dynamic scheduling with performance and energy benefits compared to the traditional schedulers because the PPO-RR model adapts to the work load profile and system conditions in a real-time fashion. Despite the fact that this study shows PPO-RR to positively perform in various metrics, no formal statistical test of significance (e.g., t-test) was conducted and will be done in subsequent research.In sum, the PPO-RR approach provides a flexible and extensible mechanism to implement sustainable cloud computing, since it contributes to satisfying the objectives of green computing with the help of well-organized schedules.

Traditional news false information detection can no longer adapt to the current information detection evasion mode. This article addressed the language changes and diversity in identifying news false information, and improved the accuracy of false information detection. Firstly, the system goal was defined to improve the accuracy of false information detection. Then, news data from FakeNewNet, BuzzFeedNews, and PoliFact platforms were collected, and the data was subjected to data cleaning, segmentation, removal of stop words, One-Hot encoding, TF-IDF (Term Frequency-Inverse Document Frequency) feature extraction, and word embedding preprocessing. After establishing a Hybrid Neural Network (HNN) model, model training, evaluation, and optimization work were carried out. In the experimental stage, 5 new datasets were added and 5 other detection algorithm models were applied to compare the detection accuracy with the proposed model. Robustness experiments were conducted to verify the robustness of the model. The experimental results showed that the accuracy of the model in this article on eight news information datasets ranged from 0.9972 to 0.9998, with a mean of 0.9987. The average accuracy of the other five algorithms was 0.8010, 0.7738, 0.7394, 0.8676, and 0.7689, respectively. The detection accuracy of the algorithm in this article was much higher than the other five algorithms, and it had high robustness. Intelligent information processing and hybrid networks have brought a more comprehensive and accurate solution to the detection of false information in news dissemination, successfully solving the problem of identifying the changing styles of false information, improving the accuracy of fake news detection, and providing a very good idea for fake news detection.

This study investigates the role of generative Artificial Intelligence (AI) in enhancing Apprenticeship Systems (ASs) by transforming Tacit Knowledge (TK) into Explicit Knowledge (EK), thereby improving Knowledge Transfer (KT) efficiency. A controlled experiment was conducted with 50 novice live-stream hosts, divided into the Experimental Group (EG) and the Control Group (CG). The EG used to train tools augmented with AI, while the CG used traditional methods. The experimental design included competency tests in seven areas, including on-camera presence, communication skills, and learning ability, and the use of statistical methods to compare the performance results of the two groups. The results established a significant improvement within the EG. The resultant indicators for expressiveness in shots (85 vs. 70), verbal expression (88 vs. 72), and learning capacity (86 vs. 71) exhibited statistically significant differences (p-values < 0.01). These outcomes suggest that the utilization of AI tools effectively enhances the development of various competencies, accelerates learning, enhances adaptability, and provides instant corrective feedback. The study's implication includes the utilization of AI in apprenticeship models, which have the potential for higher scalability, preservation of crucial TK, and workforce development, especially in industries that require Experiential Learning (EL).

Anomaly detection in modern power systems requires highly advanced techniques, as the heterogeneous data produced by power systems today can be high-volume and include sensor signals, textual logs, time series, and more. Nevertheless, conventional methods do not adapt well to dynamic data changes, lack structure, and require real-time processing, making them inefficient in more complex grids. This paper proposes a framework running on Qwen2 that leverages multi-source data fusion and cross-modal attention to address these issues. The framework can integrate sensor awareness and text logs via cross-attention, thereby aligning context and effectively finding anomalies. Additionally, it includes dynamic resolution processing to handle high-frequency sensor data and lightweight inference to support edge deployment. Experimental results on a real-world dataset show that the proposed approach delivers a 12% increase in F1-score relative to state-of-the-art models, such as Transformer-AD, and reduces false favorable rates by 50% compared with traditional methods. The framework also provides real-time footprint data with a 85 ms latency per batch, resulting in a scalable smart grid monitoring solution. The paper further develops the following practical applications of large language models in critical infrastructure by addressing deficiencies in heterogeneous data fusion, interpretability, and efficiency, given limited resources. The following plans entail minimizing energy use and scaling the system to identify multiple defect levels, such as cyberattacks and equipment wear.

As an important part of Chinese traditional culture, the innovative design of bronze patterns has always been a research hotspot in the field of computer graphics. However, traditional design methods have been unable to meet contemporary needs, so new technologies are urgently needed to promote design innovation. This article aims to propose a Transformer based Generative Adversarial Network (GAN) for bronze pattern generation, and name it Trans GAN. This paper designs a joint network architecture that combines semantic segmentation and stereo matching, uses Swin-Transformer for feature extraction, and introduces multi-scale feature fusion and CGAN framework. Through end-to-end training, the adversarial process between the generator and the discriminator is optimized to improve the diversity and realism of pattern generation. Experiments have shown that Trans GAN reduces the Fr é chet Inception Distance (FID) score to 15.2 in bronze pattern generation tasks, generation accuracy reaches 92.3%, single image generation time is 0.12 s, and artificial style score is 8.7/10. Overall, Trans-GAN shows excellent performance in bronze pattern generation and provides a new solution for the digital protection and innovative design of cultural heritage.

To analyze the development of green finance, the weights of each green finance indicator are determined by the entropy value method. To analyze the role mechanism of green financial development on economic growth and to get the relationship between the two more clearly, the study optimizes the parameters of BP network with genetic algorithm and constructs correlation analysis model. The scores of green financial development in some regions show that the development level of the Yangtze River Delta is the highest, with an average score of 62.4. The heavy industries in North China and Northeast China are well developed and have a greater demand for green financial products, with an increase of 12.7% and 11.9% respectively. The economic growth of the corresponding regions shows that the economic development level of Chang San is higher and that of the Northeast is lower. The results of the GA-BP model training show that the convergence of the algorithm is fast and the curve increases rather than decreases in the validation set, indicating that the model is well trained and the error has reached the minimum value. The mean square error of the correlation analysis model is below 0.02 and the fit coefficient R 2 is above 0.85, indicating the high accuracy of the results and the superior fit. The empirical evidence finds that the higher the level of green finance development, the better its economic development, and after a certain peak, the promotion effect diminishes. At present, some regions need to increase the proportion of green financial industries and accelerate industrial restructuring to achieve a green economy.

Selecting suitable suppliers within the cracker supply chain leads to a challenging Multi-Criteria Decision-Making (MCDM) problem due to the presence of multiple, conflicting criteria such as cost, quality, distance, and reliability. Conventional models, including the integration of Pythagorean Fuzzy AHP (PF-AHP) and Pythagorean Fuzzy VIKOR (PF-VIKOR), primarily rely on linear structures. However, these methods have shortfalls in representing the circular and non-linear nature of uncertainty in expert evaluations, which can lead to imprecise outcomes. To overcome this shortfall, this research introduces a novel MCDM approach that integrates the Circular Pythagorean Fuzzy Analytic Hierarchy Process (CPF-AHP) with the Circular Pythagorean Fuzzy VIKOR (CPF-VIKOR) method for precise decision making. CPF-AHP is employed to derive the weights of evaluation criteria through circular fuzzy approach, which captures the uncertainty and inconsistency in expert opinions in a a better way. Subsequently, CPF-VIKOR is used to rank supplier alternatives by identifying a compromise solution that considers both the best possible outcome with minimal regret. The proposed CPF-based framework effectively addresses ambiguity in expert assessments and improves the accuracy of weight determination and alternative ranking. It provides a systematic and consistent approach for aggregating expert input and demonstrates superior performance compared to traditional PF-AHP and PF-VIKOR methods, particularly in handling non-linear interrelationships among decision criteria. The integration of CPF-AHP and CPF-VIKOR enhances the decision-making process for supplier selection in uncertain and complex supply chain scenarios. By effectively managing vagueness, circular aspects, and inconsistency, the proposed model delivers a more reliable and comprehensive solution. This methodology is adaptable and can be applied to other supply chain decision problems with similar complexity and uncertainty.

With the development of the Internet and social media, the spread speed and breadth of fake news have significantly increased, which has brought severe negative impacts to society. Existing false information detection algorithms often lack robustness and generalization when dealing with constantly changing, complex news content. The problems with traditional research mainly include limited training data, insufficient robustness of detection algorithms, and poor generalization ability in diverse environments. This article first used GAN (Generative Adversarial Network) to generate realistic fake news data and expand the training set; secondly, a virtual reality environment was constructed to simulate users’ news consumption experiences in different contexts and collect user interaction data; then, combining Text, images, and user interaction data, deep learning models such as long short-term memory networks and convolutional neural networks were used to extract multimodal features; finally, the multimodal features were input into the fusion detection model for comprehensive analysis and decision-making. Through these steps, this article achieved significant detection results in a complex and ever-changing news environment. The research results indicated that the method proposed in this article not only improved the accuracy of fake news detection but also enhanced the model's adaptability across different contexts. The detection model's accuracy increased by 12%, from 80% to 92%. The fusion of virtual reality and generative adversarial networks significantly improved the detection model's robustness across diverse news contexts, with strong application prospects and practical value.

Car classification, using different machine learning models with optimization frameworks, is done for evaluation in this work. We used different models, namely, Extra Trees, XGBoost, Gaussian Naive Bayes, K-Nearest Neighbors, Histogram-based Gradient Boosting (Hist Gradient Boosting), and Linear Discriminant Analysis, for classification. The car samples are classified as “very good,” “good,” “acceptable,” and “unacceptable.” Among these, Hist Gradient Boosting has the highest value for precision, accuracy, recall, and F1 score. We further tuned this model using Evolutionary Strategies, Evolutionary Programming, Covariance Matrix Adaptation Evolution Strategy, and the Flower Pollination Algorithm. Our outcomes indicate that the Covariance Matrix Adaptation Evolution Strategy and Flower Pollination Algorithm significantly enhanced the performance of the model and outperformed Evolutionary Programming. This work investigates the potential of integrating advanced machine learning models with sophisticated optimization strategies to deliver an effective car evaluation classification process that will be useful in this industry and, perhaps, in many other classification tasks.

With the popularity of digital advertising, the news media have presented new characteristics in reporting on the development of renewable energy resources. Through data analysis, we found that news reporting on renewable energy has experienced significant growth in digital advertising. News coverage of renewable energy in digital advertising increased by 75% between 2018 and 2023, which indicates the growing interest in the industry. Moreover, these reports often contain a wealth of data and facts, providing readers with detailed information. Five main data points generally exist for reports, which refer to aspects such as development, energy use, technological change, etc. In digital advertising, renewable energy news reports also have the characteristics of cross-regional communication. According to the data, these reports cover more than 100 countries and regions around the world, effectively promoting exchanges and cooperation in different regions. At the same time, these studies typically focus on the advantages of renewable energy, encouraging the public to become advocates for policies that can influence legislation, capital investment, and the industry's sustainable growth. Following their review of these articles, 70% of survey respondents in a recent study indicated an increased favourable attitude toward renewable energy. In conclusion, with a focus on digital advertising, renewable energy news reports play a crucial role in shaping the public's perceptions and supporting the development of the sector. They not only provide a wealth of data to support them but also have a broad international perspective and positive emotional tendencies. These features help to drive sustainable development in the renewable energy sector and stimulate public attention and support for the sector.

With high penetration of renewable energy in power system, the demand side response is gradually considered to be an important flexible resource, leading to active participation in electricity market. This article provides a design of demand response trading platform for load aggregators, with integration of end-edge-cloud collaboration and blockchain. Firstly, the end-edge-cloud collaborative architecture enables distributed resources entering electricity market efficiently and realizes the privacy protection of users’ production and operation data, by hierarchical decomposition and coordination of computing and data storage tasks. Meanwhile, certifying and accounting the transaction results through Consortium Blockchain technology can improve security and credibility of transaction on the Platform. Secondly, credit evaluation is taken into account when distributed resources aggregation and instructions decomposition, which encourages the user to strictly execute the contract, thus, mitigate the negative impact for aggregators. Credit evaluation is only carried out with the help of consortium blockchain, providing a secure, efficient and mutually trusted trading ecosystem.

Neuromuscular disorders consist of a broad range of disease that affects muscles and nerves, which often lead to significant disability. To effectively manage these diseases, early diagnosis is crucial, but gold standard techniques such as electromyography and genetic testing are invasive, time-consuming, and costly. However, ultrasound, being non-invasive and cost-efficient can be a reliable alternative, but manual interpretation is a major limitation of this method. Our study presents a deep learning-based approach to automate the diagnosis of neuromuscular disorders.

Our study leveraged Convolutional Neural Networks (CNNs) for feature extraction, combined with dense network for the classification task. To address the black-box nature of deep learning models, Local Interpretable Model-Agnostic Explanations (LIME) was employed, which segments that part of ultrasound images that influence the prediction. The proposed model achieved an accuracy of 94% with and a ROC-AUC score of 0.97, an F1 score of 0.80, a precision of 0.83 and a recall of 0.77 which demonstrates the ability of model to accurately differentiate between the healthy and pathological muscle ultrasound images. The interpretability mechanism (LIME) was successfully able to segment those part of ultrasound images that were influencing the model's prediction giving insights into prediction mechanism and enhancing trust on model. Our approach eliminates the need of image segmentation and need for manual feature extraction by employing end to end Convolutional Neural Network (CNN) model achieving 94% accuracy with balanced precision and recall. Also, we employed explainable mechanism which enhances the trust on the model which make it more adaptable for clinical applications. Our study presents a promising deep-learning based alternate for the diagnostic of neuromuscular disorder using ultrasound images with high accuracy.

The integration of an interpretability mechanism enhances the trust on model making it suitable for integration in clinical applications. By automating the diagnostic mechanism neuromuscular disorders can be detected early which will help in strategize the management of neuromuscular disorders eventually enhancing the patient health.

Extreme Value Theory, or univariate EVT, is widely used to assess structural risks of failure or damage, brought on by excessive environmental stressors in structural design. In engineering practice, a combination of multiple cross-correlated system components and covariates, rather than a single, univariate load, is what causes failure or damage. A multimodal state-of-the-art reliability-based approach for the multivariate structural design is presented. State-of-the-art pre-asymptotic multivariate methodology in combination with an accurate extrapolation scheme was utilized to model the Joint Probability Distribution Function (JPDF) tail of an M-dimensional random/stochastic process. The primary aim of this study was to envisage a generic, state-of-the-art multivariate reliability approach for assessing the failure or damage risks of high-dimensional dynamic systems. Note that for a multidimensional series-type system, its failure is given by a first passage event, and any parallel-type system can be equivalently reformulated as a series-type one.

Novelty: The advocated multidimensional structural reliability approach would enable the extraction of relevant excessive dynamics information from time histories that had been physically recorded or numerically simulated. A variety of multimodal nonlinear dynamic systems can have their failure (damage) risks accurately and efficiently predicted using the proposed multimodal hypersurface Gaidai reliability methodology, which considers non-stationarity and memory (clustering) effects. High-dimensional, big data, deep-sea, ocean engineering and aerospace applications can benefit from the advocated multidimensional reliability approach.

With the acceleration of green transformation and digitalization, the integrated development of the green digital economy and the new energy industry has become the core driving force for promoting regional coordination and high-quality development. Based on the macro perspective of integrated development, this paper constructs a multi-agent game model of the coordinated evolution of the green digital economy and the new energy industry. It systematically describes the mechanisms of interactive behavior among the government, enterprises, and platform entities, driven by resource allocation, technical collaboration, and policy incentives. The model simulation results reveal that policy incentive variables are prone to causing economic pressure or regional imbalance. Based on the analysis path and research, it is shown that when the PCI (Policy Coordination Index) is 0.9, the growth rate of digital economy output value is 8.4%, and the coordination promotion factor is 2.27. The average TSS (Technology Sharing Rate) of cities with technology-sharing mechanisms is 0.655; the average FFI (Factor flow index) of the technology-support optimization group is 0.713. Improving the effectiveness of policy linkage, promoting technology-sharing mechanisms, and optimizing the flow of factors between regions are key paths to achieving deep integration between the green digital economy and new energy industries. This paper provides methodological support for the theory of integrated development, a theoretical basis, and strategic inspiration for promoting regional coordination and for formulating green transformation policies.

A novel Life Cycle Assessment (LCA) is presented in this study to model carbon emission reductions, specifically in the construction of a power grid that favours the integration of low-carbon technologies. Unlike traditional assessments, this research alone assesses emissions from the whole construction process through planning, material procurement, transportation, installation and the operation phases of the building; and does it uniquely by comparing emissions of traditional versus energy efficient systems, smart grid implementations, low carbon materials and electric vehicles. Flow through more combinations of low-carbon technologies and more plant system cases are found to result in a significant overall emission reduction of 17.5%. Furthermore, the study compares it with existing models, such as Environmental Impact Assessment (EIA) and Carbon Footprint Analysis (CFA), and finds that life cycle analysis is a more efficient and accurate method for identifying emission-reduction potential. Therefore, this study offers an effective methodological blueprint for policymakers to develop regulations and incentives specifically tailored to incentivise sustainable construction practices and achieve global climate change mitigation and carbon neutrality.