Subgraph-level federated graph neural network for privacy-preserving recommendation with meta-learning.

Graph neural networks (GNN) are widely used in recommendation systems, but traditional centralized methods raise privacy concerns. To address this, we introduce a federated framework for privacy-preserving GNN-based recommendations. This framework allows distributed training of GNN models using local user data. Each client trains a GNN using its own user-item graph and uploads gradients to a central server for aggregation. To overcome limited data, we propose expanding local graphs using Software Guard Extension (SGX) and Local Differential Privacy (LDP). SGX computes node intersections for subgraph exchange and expansion, while local differential privacy ensures privacy. Additionally, we introduce a personalized approach with Prototype Networks (PN) and Model-Agnostic Meta-Learning (MAML) to handle data heterogeneity. This enhances the encoding abilities of the federated meta-learner, enabling precise fine-tuning and quick adaptation to diverse client graph data. We leverage SGX and local differential privacy for secure parameter sharing and defense against malicious servers. Comprehensive experiments across six datasets demonstrate our method's superiority over centralized GNN-based recommendations, while preserving user privacy.

Multi-objective optimization methodology for green-gray coupled runoff control infrastructure adapting spatial heterogeneity of natural endowment and urban development.

Cost-effective runoff control scheme drafting involves localization, multi-sector coordination, and configuration of multifunctional infrastructures. Numerous independent variables, parameters, weights, and objectives make runoff control optimization quantitatively arduous. This study innovatively proposed a multi-objective optimization methodology for green-gray coupled runoff control infrastructure adapting spatial heterogeneity of natural endowment and urban development. The quantitative methods of multi-objective evaluation, hydrological feature partition, and pressure-adapted multi-objective weight assignment were proposed. Remote sensing inversion of water quality, hydrological model simulation (using SWAT and SWMM software), landscape pattern index calculation, life cycle cost (LCC), life cycle assessment (LCA) on ecological impact, and NSGA-II optimization algorithm were applied. Wuhan, the most water-sensitive city in China, was studied as a case. Runoff control function (RCF), capital investment (CI), and ecological return on investment (EROI) served as optimized objectives. High, medium, and low built-up regions in Wuhan urban development planning district were extracted by topographic factors and landscape patterns, which comprised 28, 34, and 38% of the case area, respectively. Three corresponding hydrological models were then built to illustrate distinct runoff control cost-efficiency in each region. Pressure distributions on runoff control, economic constraints, and ecological resource scarcity were quantitatively evaluated. And four pressure zones were clustered, which occupied 36, 29, 16, and 19% of the case area, respectively. Then the zonal weighted optimization decision-making matrix (with 3 hydrological models and 5 wt) was established by overlaying the pressure zone and built-up zone. In high, medium, and low built-up regions, optimized solutions reduced annual runoff volume by 86, 82%, and 77%The average runoff investments per square meter of impervious underlying surface in high, medium, and low built-up regions were 34.2, 18.7, and 7.9 RMB yuan, respectively. Medium and low built-up regions may only need 55 and 23% of the high built-up region for the unitary impervious underlying surface to balance runoff control and ecological benefits. Runoff control and financial utilization efficiency enhance with hydrological differentiation zones. Thus, the optimization solutions are zonal adaptive, refined, comparable, replicable, and implementable.

Human physiological responses of exposure to extremely cold environments.

Extremely cold events have occurred more frequently in the past few years. People exposed to extremely cold exposure could suffer the threats of human health and safety like cold stress and injury. This study aims to investigate human physiological responses of exposure to extremely cold environments and the moment of temperature step. The experiments of 12 subjects exposed to three different cold exposure conditions (-5 °C, -10 °C, -15 °C) were carried out in a climate chamber. Most critical physiological parameters, including the core temperature, local skin temperature, blood pressure, heart rate, respiration rate and blood oxygen saturation, were measured to evaluate human physiological responses. In the particular short term study, the results show that the local skin temperature and blood pressure are the most significant indexes for evaluating the risk of cold strain in extremely cold environment. The finger temperature is a critical index of hand and finger flexibility, and it will lead to serious injuries and reduced manual performance when exposed to below -5 °C for more than 20 min. The high physiological strain at the very beginning moment of cold exposure can significantly affect the ability to make correct judgment and action, and it is suggested that the personnel adapt for 3 min after entering into the extremely cold environment to stabilize physiological parameters and thus enhancing the safety and occupational performance. The experimental data of this study is also of great significance for the development and validation of thermophysiological models.

Framework Design and Influencing Factor Analysis of a Water Environmental Functional Zone-Based Effluent Trading System.

The efficacy of traditional effluent trading systems is questionable due to their neglect of seasonal hydrological variation and the creation of upstream hot spots within a watershed. Besides, few studies have been conducted to distinguish the impacts of each influencing factor on effluent trading systems outputs. In this study, a water environmental functional zone-based effluent trading systems framework was configured and a comprehensive analysis of its influencing factors was conducted. This proposed water environmental functional zone-based effluent trading systems was then applied for the control of chemical oxygen demand in the Beiyun River watershed, Beijing, China. Optimal trading results highlighted the integration of water quality constraints and different hydrological seasons, especially for downstream dischargers. The optimal trading of each discharger, in terms of pollutant reduction load and abatement cost, is greatly influenced by environmental and political factors such as background water quality, the location of river assessment points, and tradable discharge permits. In addition, the initial permit allowance has little influence on the market as a whole but does impact the individual discharger. These results provide information that is critical to understanding the impact of policy design on the functionality of an effluent trading systems.

Uncertainty analysis for an effluent trading system in a typical nonpoint-sources-polluted watershed.

Conventional effluent trading systems (ETSs) between point sources (PSs) and nonpoint sources (NPSs) are often unreliable because of the uncertain characteristics of NPSs. In this study, a new framework was established for PS-NPS ETSs, and a comprehensive analysis was conducted by quantifying the impacts of the uncertainties associated with the water assimilative capacity (WAC), NPS emissions, and measurement effectiveness. On the basis of these results, the uncertain characteristics of NPSs would result in a less cost-effective PS-NPS ETS during most hydrological periods, and there exists a clear transition occurs from the WAC constraint to the water quality constraint if these stochastic factors are considered. Specifically, the emission uncertainty had a greater impact on PSs, but an increase in the emission or abatement uncertainty caused the abatement efforts to shift from NPSs toward PSs. Moreover, the error transitivity from the WAC to conventional ETS approaches is more obvious than that to the WEFZ-based ETS. When NPSs emissions are relatively high, structural BMPs should be considered for trading, and vice versa. These results are critical to understand the impacts of uncertainty on the functionality of PS-NPS ETSs and to provide a trade-off between the confidence level and abatement efforts.

[Research on the virtual water composition and virtual water trade for agriculture in Beijing].

Based on the irrigation norm of typical district and county, and revised by the isoline map of Chinese crops water demand, the change of crops program was analyzed as well as the agricultural water use and its GDP benefits. Then the virtual water was calculated for years. At last, the input-output method was used to calculate the trade of virtual water in Beijing. As the results, the virtual water for cereal crops has been decreasing in Beijing, from 1.832 x 10(9) m3 in 1990 to 4.283 x 10(8) m3 in 2004. Otherwise the virtual water for technical crops has been increasing, which is from 9.06 x 10(8) m3 in 1990 to 1.492 x 10(9) m3 in 2004. On the whole, the virtual water for crops has been decreasing in Beijing. From the angle of primary products Beijing is a virtual water importing area. Virtual water importing of annual average is 2.37 x 10(8) m3, which is about 5.93% of the total water of Beijing. Virtual water has been an important supplement of local real water of Beijing.

[Distribution of virtual water of crops in Beijing].

Virtual water content of grains and vegetables in Beijing's districts is calculated and analyzed for many years by irrigating water quota method, which is compared with the distribution and exploitation of groundwater in Beijing. The results indicate the virtual water content of grains shows a downward trend in all the districts, but the grain production in Yanqing district brings great pressure to the local groundwater. Secondly, the virtual water content of vegetables shows an upward trend in Shunyi District, Daxing district and Pinggu District and is accounting for more and more gradually. Thirdly, the total virtual water volume of grains is decreasing, and the total virtual water volume of vegetables is increasing and the total virtual water volume of crops in Beijing is reducing in recent years, which corresponds with the structural adjustment of policies.