This paper presents a data-efficient object detection framework that integrates YOLO with few-shot learning techniques to mitigate the challenges of large-scale annotated data dependency and small object detection. By incorporating Feature Pyramid Networks (FPN) and spatial attention mechanisms, the framework enhances detection accuracy for small objects. Additionally, the use of few-shot learning approaches — meta-learning, data augmentation, and transfer learning — enables the model to generalize effectively from limited data while preserving real-time inference speed. Experimental results demonstrate that the proposed framework excels in data-scarce scenarios, making it suitable for applications such as autonomous driving, aerial surveillance, medical imaging, and wildlife monitoring. Future research will focus on optimizing computational efficiency, enhancing cross-domain adaptability, and exploring advanced few-shot learning strategies. This work provides a scalable and effective solution for object detection in resource-limited environments.

Leveraging computer systems and information technologies, intelligent manufacturing demonstrates enhanced collaborative capabilities and elevated decision-making capacity. From a digital cognitive framework and information system-oriented view, this paper focuses on the process of supply chain digital transformation, interpreting the methodology and approaches to eliminate obstacles such as the lack of data standards and inaccurate data analysis, lack of supply chain resilience, contradiction between customized and operational efficiency in production systems, uncooperative market conditions and other internal or external factors. To address these challenges, the systematic integration of big data analytics, deep learning-based intelligent optimization, and blockchain-enabled collaborative mechanisms establishes practical technical pathways. This convergence augments supply chain cognitive capabilities, advances intelligent production scheduling competencies, amplifies market adaptive intelligence, and facilitates sustainable evolution of intelligent supply chain ecosystems.

In view of the problems of complex housing information in the real estate market and the difficulty for users in finding houses, the personalized recommendation system of housing information based on deep learning is studied. The system adopts a hierarchical architecture. The data layer collects multi-source data from the housing information platform and stores it in MySQL and MongoDB databases respectively according to the structural characteristics. The preprocessing layer improves the data quality by removing noise, processing stopped words and eliminating duplicate data. The recommendation layer uses the convolutional neural network to generate the recommendation results. The service layer realizes invalid housing filtering, grouping and sorting optimization, and recommendation reason adding functions. The experiment shows that the recommendation reliability of the system is high, and significantly improves the efficiency of user house hunting, greatly shortens the house search and transaction time, and has a good effect of house information recommendation.

In recent years, with the continuous development of cloud computing technology, the importance of cloud computing data center, as the infrastructure of cloud computing has become increasingly prominent. Cloud computing data centers need to meet the requirements of high efficiency, high availability, scalability, security and other aspects to ensure the stable operation of cloud computing services and a good experience for users. This paper first introduces the basic concept and development status of cloud computing data center, then analyzes the key elements and objectives of cloud computing data center architecture design, and then proposes an efficient cloud computing data center architecture design scheme, and the effectiveness and performance of the scheme are verified through experiments. The experimental results show that this scheme can improve the resource utilization rate of cloud computing data center, reduce energy consumption, improve the reliability and scalability of the system, and has certain practical application value.

Compared to symmetric cascaded H-bridge inverters (CHB), hybrid cascaded inverter technology can generate more voltage levels with fewer inverter units. In traditional inverter modulation strategies, hybrid cascaded inverters produce high-quality voltage output. However, due to power imbalance among cascaded units, the uneven power distribution accelerates unit degradation, thereby affecting the overall lifespan of the inverter. To address these issues, this paper proposes a novel modulation strategy. The low-voltage units undergo carrier reconstruction, while the high-voltage units employ modulation wave reconstruction, achieving power balancing across all units. The low-voltage units exhibit improved power-balancing response speed compared to traditional methods. Simulation results validate the effectiveness of the proposed approach.

Urban flooding disasters frequently occur in our country, severely affecting the national development process, anticipating the probability and severity of floods can effectively reduce the negative impacts caused by floods, the rapid progress of hydrology has accelerated the development of flood prediction research. Currently, a lot of machine learning methods are widely applied in the field of flood forecasting based on hydrology, which holds great significance for social development. First, the hydrological models currently used for flood forecasting are introduced. Then, the application of machine learning models in hydrology is elaborated. Finally, the problems and challenges faced by machine learning in flood prediction are analyzed and summarized, and prospects for future flood prediction technologies are discussed.

Currently, addressing the imbalances in the development of domestic general air freight and public aviation freight formats, as well as the imperfections in policies, regulations, and sys-tems, is crucial. This study employs system dynamics theory as its research method to con-struct a dynamic model of China's general aviation freight system. By examining the supply side, demand side, and the general aviation freight industry chain, the study illustrates the feedback relationships within general aviation freight and establishes an indicator system. Using Ven sim PLE software, the research creates causal relationship diagrams of the main subsystems and an overall system flow diagram. The model is quantitatively specified through historical and sensitivity testing. The findings indicate that varying levels of financial in-vestment and tax preferential policies significantly enhance the economic benefits of navigation and freight in China. Additionally, increased investment in scientific research funding is expected to advance the technical level of navigation operations. Therefore, it is recommended to boost investment in general air freight, implement preferential tax policies for air-craft manufacturing enterprises, navigation operating enterprises, and infrastructure support units, increase investment in R&D for the general cargo transport industry, and focus on the training and incentivization of navigation professionals.

The new framework that combines the low-altitude economy with high-speed transportation offers a practical and effective solution for modern city logistics systems. It shows strong potential in boosting delivery efficiency, cutting costs, and reducing carbon emissions. This study focuses on Suzhou's logistics network, using a multi-goal optimization model and a smart scheduling system to assess how well the low-altitude economy works in different logistics situations. By combining an ARIMA time series prediction model with a Long Short-Term Memory (LSTM) network, the study looks at how logistics needs change during busy times, bad weather, and emergency situations. The research uses different data sources to help drones and ground vehicles work together smoothly. The model includes three main measures: delivery efficiency (T), transportation costs (C), and carbon emissions (E), and shows the real benefits of the low-altitude economy through clear data analysis. The results show that in normal conditions, delivery efficiency increased to 97.9, transportation costs dropped to 65.4, and carbon emissions fell to 58.2. During peak traffic and bad weather, delivery efficiency stayed strong at 85.5 and 80.3. The smart scheduling system managed resources well, keeping costs and emissions within safe limits (cost: 70.1-73.5; emissions: 61.8-67.9). Scheduling efficiency went up from 0.85 to 0.93, and resource use improved from 74.6% to 88.1%. The analysis showed a clear negative link between delivery efficiency and carbon emissions (-0.85) and a positive link between costs and emissions (0.78). This suggests the need to balance cost savings with environmental benefits. The suggested approach, combining the multi-goal optimization model with a smart scheduling system, not only helps Suzhou handle complex logistics challenges but also provides a useful model for other large cities, supporting the move towards faster, greener, and smarter city logistics systems.

This paper explores the role of AI-assisted strategies in enhancing Chinese language proficiency for non-native students preparing for the AP Chinese exam. By analyzing the key requirements of the AP Chinese exam, the study identifies how AI tools, such as Speechling, FluentU, and HelloChinese, can support students in improving their listening, speaking, reading, and writing skills. The paper also addresses the ethical and social implications of AI in education, focusing on data privacy, algorithmic bias, and educational equity. Key strategies for optimizing AI in language learning are discussed, alongside a comparison of different AI tools in AP exam preparation. The study emphasizes the need for a balanced approach to AI implementation in education, considering both the technological advantages and the ethical challenges. Future research directions are outlined to further explore the long-term impact and integration of AI in educational contexts.