With the development of artificial intelligence technology, text-driven image generation has gradually become a research hotspot. In terms of the application of traditional Chinese painting generation, due to the particularity of traditional Chinese painting themes, techniques, and artistic conception, existing models have problems such as cross-modal alignment deviation. To enhance the model's understanding of text semantics and improve the matching degree between text and generated Chinese painting images, a multimodal dataset of Chinese paintings is studied, and the Vector Quantized Generative Adversarial Network (VQGAN) is improved. A new multimodal text generation method is constructed by combining transformer-based bidirectional encoder representation, Transformer decoder, and contrastive language-image pre-training models. The results showed that when trained for 100 rounds on the general object dataset and Flickr30k dataset in context, the Inception Score (IS) values of this method were 39.6 and 5.1, respectively, and the Fréchet Inception Distance (FID) values were 2.3 and 1.7, respectively, which were better than models such as VQGAN. In the ablation experiment, the IS was 5.8, the FID was 10.3, and the cosine similarity was 0.82. The convergence was achieved after 60 rounds, which was better than other variants. The method proposed in the study has good adaptability to the generation of traditional Chinese painting. Although the expansion of complex scenes is slightly weak, the overall performance is excellent, filling the gap in traditional artistic semantic mapping and providing support for the digital dissemination and innovation of traditional Chinese painting.

To overcome the challenges of decoding pattern genes and the lack of innovative designs in the digital transformation of Miao embroidery as an intangible cultural heritage, a deep learning fusion architecture based on a convolutional neural network Transformer (CNN Transformer) is proposed. This architecture aims to accurately analyze the semantics and generate stylized Miao embroidery patterns. This architecture uses a ResNet-50 backbone network to extract local pattern details and combines an improved atrous spatial pyramid pooling (ASPP) module to preserve edge structures for semantic segmentation. It simultaneously utilizes a progressive Transformer encoder and multi-scale channel attention to dynamically focus on the pattern subject, fuse features, and achieve style transfer. The experimental results showed that the segmentation accuracy of the model on four types of embroidery patterns exceeded 85% (up to 92.68%), and the structural similarity (SSIM) reached 0.957. In the style transfer task, the style fidelity index (KL divergence 0.52, Gram matrix MSE 0.019) and average accuracy (AP 0.926) of this method were significantly better than the comparison algorithms. This research method effectively solves the problems of inaccurate segmentation of complex patterns and distortion of cultural symbols in traditional style transfer methods. It provides a technical reference for analyzing and innovatively activating Miao embroidery culture.

This research investigates the impacts of applying Green Building (GB) principles within Jordanian construction firms, focusing on their effects on environmental, economic, and social dimensions. A descriptive analytical method was employed, suitable for social and humanitarian research contexts. The study targeted a sample of 15 large construction companies listed on the Amman Stock Exchange, utilizing a random sample of 150 individuals, including heads of departments, engineers, designers, architects, supply chain managers, and directors. Data collection was conducted through questionnaires, with 150 distributed and 110 valid responses received, resulting in an 80% response rate. Data analysis was performed using SPSS software version 22 to calculate means, frequencies, and standard deviations. The findings revealed significant impacts of applying GB principles, with a correlation coefficient of 0.754 for environmental quality, indicating that GB practices account for 56.8% of the variance in environmental outcomes. For residents' health, the correlation was 0.643, explaining 41.3% of the variance, while resource preservation showed a strong correlation of 0.749, indicating substantial contributions. Economically, the principles demonstrated a correlation of 0.705, accounting for 57.3% of the variance in economic performance. These findings underscore the necessity of integrating GB practices into construction projects to enhance sustainability, with recommendations including early integration of green design in project development, establishment of comprehensive green education programs, provision of incentives for existing building owners, and securing funding for renewable energy initiatives. Implementing these strategies is crucial for maximizing the effectiveness of GB practices and advancing sustainable development in Jordan.

Over time, intrusion detection systems have grown essential in ensuring network security by identifying malicious activities within network traffic and alerting security teams. Machine learning techniques have been employed to develop these systems. However, these approaches often face challenges related to low accuracy and high false alarm rates. Deep learning models like Long Short-Term Memory (LSTM) are utilized to address these limitations. Despite their potential, LSTM models require numerous iterations to achieve optimal performance. This study introduces an enhanced version of the LSTM algorithm, termed ILSTM, which integrates the Salp Swarm Optimizer (SSO) to boost accuracy. The ILSTM framework was applied to construct an advanced intrusion detection system capable of binary and multi-class classifications. The approach comprises two phases: The first involves training a standard LSTM model to initialize its weights. In contrast, the second employs the SSO hybrid optimization algorithm to fine-tune these weights, enhancing overall performance. The effectiveness of the ILSTM algorithm and the intrusion detection system was assessed using two publicly available datasets, NSL-KDD and LITNET-2020, across nine performance metrics. Results demonstrated that the ILSTM significantly outperformed the conventional LSTM and other comparable deep learning models in accuracy and precision. Specifically, the ILSTM achieved an accuracy of 93.09% and a precision of 96.86%, compared to 82.74% accuracy and 76.49% precision for the standard LSTM. Moreover, the ILSTM exhibited superior performance on both datasets and was statistically validated to be more robust than LSTM. Furthermore, the ILSTM excelled in multiclass intrusion classification tasks, effectively identifying intrusion types.

Objectives: This paper aims to explore how agricultural product sales via live streaming affects the purchasing behavior in the context of the digital economy and evaluate the correlation between them. Methods: A questionnaire was designed to collect respondent’ personal information and information on their purchases of agricultural products. The correlation between agricultural product purchases and the live-streaming economy was measured. Findings: Most respondents purchased agricultural products on platforms such as Douyin and Taobao, preferred watching live streaming of internet celebrities and farmers, primarily bought fruits, vegetables, whole grains, and coarse cereals, and expressed high satisfaction with their agricultural product purchases. Correlation analysis indicated that the correlation coefficient between agricultural products and purchases in the live-streaming economy was highest at 0.742. Regression analysis found a significant positive correlation between agricultural products, anchors, live streaming, platforms, and agricultural product purchases. Novelty: The research quantifies the relevant information on agricultural product live streaming and purchases through questionnaire analysis. It also reveals the positive influence of the digital economy on agricultural product purchases, providing some references for the further development of agricultural product live streaming.

This paper proposes a data-driven adaptive scheduling method based on the Deep Deterministic Policy Gradient (DDPG) algorithm to address the challenges that traditional vehicle dynamic maintenance scheduling methods struggle to cope with real-time, complex and resource optimization issues. A mathematical model of vehicle dynamic maintenance scheduling is constructed, defining the state space, action space and reward function. Then, the DDPG reinforcement learning framework is used to optimize strategies through the Actor-Critic structure. Contrastive experiments are also carried out in a simulation environment to evaluate the algorithm's performance. The results indicate that the DDPG algorithm achieves an average maintenance response time of 23.4 minutes, approximately 34% shorter than the genetic algorithm. Its resource utilization reaches 88.7%, over 13% higher than traditional methods. Moreover, the maintenance satisfaction score is 4.6 out of 5. The findings show that the algorithm has remarkable advantages in multi-objective scheduling optimization and provides feasible paths and technical support for the intelligence of vehicle dynamic maintenance systems.

To better monitor and characterize sounds produced by substations, this study aims to separate sounds produced by the equipment from environmental ambient noise as a means of improving the relevancy (and ultimately reliability) of the power grid. To do so, we propose a deep learning-based noise monitoring system in an end-network-cloud architecture that enables remote data collection, analysis, and management. This is achieved by developing a deep learning-based noise monitoring system, enabling remote data collection, processing, and management. The proposed method consists of two basic components: a self-designed Panel Response Acquisition device that can collect sufficient acoustic information, and a refined Deep Belief Network (DBN) that is trained with a Dynamic version of the Dwarf Mongoose Optimizer (DDMO) to improve the accuracy of the noise separation process. The performance of the DBN/DDMO model is 13.1 dB for SI-SDRi and 15.7 dB for SDRi, which are large improvements for SI-SNRi and SDRi over AlexNet and CNN-VGG19. This approach minimizes SPL deviations, as shown by a thorough computation regarding several data sets; therefore, it guarantees precise noise quantification under disturbing sounds. By allowing for proactive identification of unusual noise levels, this research supports predictive maintenance methods that can avoid sudden failures and improve the overall reliability of substations.

The deepening of globalization has posed challenges to open economies, such as fluctuations in international capital flows and intensified cross-border risk contagion. To explore the impact mechanism of FC on MP and ERP, this paper adopts TVP-VAR, MS-VAR, and MS-DSGE models, and introduces the SVR model as an auxiliary prediction tool to analyze policy dependency characteristics through standardization and periodic decomposition. The results showed that during the 2008 financial crisis, the growth rate of the broad money supply reached 17.0%-20.0%, the Shanghai Interbank Offered Rate rose to 3.6%-5.2%, and the asset price volatility exceeded 20%. During the COVID-19 pandemic in 2020, the volatility of real estate prices reached 7.2%-9.5%. In terms of policy transmission, the impact of asset price shocks on the consumer price index significantly increased after 3 months and reached its peak after 6 months. The regulatory coefficient of interest rate policy on the financial condition index under the high volatility regime was 1.1862, and the response coefficient of the growth rate of the broad money supply to the output gap under the low volatility regime was 0.2156. The SVR model had a prediction accuracy (R2 of 0.85) for the impact of MP on ERVs, especially in capturing nonlinear relationships during financial expansion periods. This achievement demonstrates the significant effect of FC stages on the effectiveness of MP, providing an FC sensitive policy framework for open economies, helping to enhance macroeconomic resilience and maintain internal and external balance.

This research develops a comprehensive framework for optimizing green business information management systems to achieve carbon neutrality goals through digital transformation. The study conducted cross-sector carbon footprint assessments of information systems across six industries, analyzing emission patterns based on operational scales, industry characteristics, and technological architectures. A multi-tiered optimization model was developed targeting infrastructure, data management, and application layers, validated through empirical data from enterprises undergoing digital transformation. Results reveal a strong negative correlation (r = -0.73) between digital maturity indices and emission intensity, with organizations implementing comprehensive digital transformation achieving average carbon reductions of 31% over five years. The proposed multi-tiered optimization approach enabled 42.6% emission reductions, with technology companies achieving 68% reductions. Economic analysis demonstrates return on investment ranging from 132-278% over five-year periods, with payback periods of 14-36 months. This study advances information management theory by integrating technological architecture with environmental performance governance, providing quantifiable carbon assessment methodologies across system layers and practical implementation matrices for industry-specific applications. The framework enables organizations to balance carbon reduction objectives with operational efficiency, addressing the critical gap between theoretical potential and practical implementation in carbon-neutral transformations.

The research aims to enhance the effectiveness of the public opinion evolution guidance model, with a particular focus on the influence of opinion leaders, and address the shortcomings of traditional models, such as neglecting opinion leaders and insufficient network topology. Therefore, the relevant scale-free network is used to improve the traditional public opinion evolution model. An improved model integrating the two is proposed by introducing a real coded accelerated genetic algorithm. The experimental results show that the proposed model converges to four opinion clusters, with the average values of negative and positive opinions being 0.399 and 0.370, respectively, demonstrating the trend closest to the actual data. When the parameters are fixed, the ultimate development of public opinion shows obvious changing trends in different situations, and the validity of the model has been proved by practice. The research innovatively introduces the scale-free network based on Barabasi and Albert, and improves the Hegselmann Krause model. Meanwhile, by comprehensively considering the influence of opinion leaders and network topology factors, the model overcomes the shortcomings of traditional models in public opinion guidance and also demonstrates good practicability in practical applications.