Multicenter Hierarchical Federated Learning With Fault-Tolerance Mechanisms for Resilient Edge Computing Networks.

In the realm of federated learning (FL), the conventional dual-layered architecture, comprising a central parameter server and peripheral devices, often encounters challenges due to its significant reliance on the central server for communication and security. This dependence becomes particularly problematic in scenarios involving potential malfunctions of devices and servers. While existing device-edge-cloud hierarchical FL (HFL) models alleviate some dependence on central servers and reduce communication overheads, they primarily focus on load balancing within edge computing networks and fall short of achieving complete decentralization and edge-centric model aggregation. Addressing these limitations, we introduce the multicenter HFL (MCHFL) framework. This innovative framework replaces the traditional single central server architecture with a distributed network of robust global aggregation centers located at the edge, inherently enhancing fault tolerance crucial for maintaining operational integrity amidst edge network disruptions. Our comprehensive experiments with the MNIST, FashionMNIST, and CIFAR-10 datasets demonstrate the MCHFL's superior performance. Notably, even under high paralysis ratios of up to 50%, the MCHFL maintains high accuracy levels, with maximum accuracy reductions of only 2.60%, 5.12%, and 16.73% on these datasets, respectively. This performance significantly surpasses the notable accuracy declines observed in traditional single-center models under similar conditions. To the best of our knowledge, the MCHFL is the first edge multicenter FL framework with theoretical underpinnings. Our extensive experimental results across various datasets validate the MCHFL's effectiveness, showcasing its higher accuracy, faster convergence speed, and stronger robustness compared to single-center models, thereby establishing it as a pioneering paradigm in edge multicenter FL.

How skillful was the projected temperature over China during 2002-2018?

Climate change has attracted significant attention due to its increasing impacts on various aspects of the world, and future climate projections are of vital importance for associated adaptation and mitigation, particularly at the regional scale. However, the skill level of the model projections over China in the past more than ten years remains unknown. In this study, we retrospectively investigate the skill of climate models within the Third (TAR), Fourth (AR4), and Fifth (AR5) Assessment Reports of the Intergovernmental Panel on Climate Change (IPCC) for the near-term projections of near-surface (2 m) air temperature changes in China. Those models are revealed to be skillful in projecting the subsequent climatology and trend of the temperature changes in China during 2002-2018 from several to ten years ahead, with higher scores for the climatology than for the trend. The model projections display cold biases against observations in most of China, while the nationally averaged trend is overestimated by TAR models during 2002-2018 but underestimated by AR4 models during 2008-2018. For all emission scenarios, there is no obvious difference between the equal- and unequal-weighted averages based on the arithmetic averaging and reliability ensemble averaging method respectively, however the uncertainty range of projection is narrowed after weighting. The near-term temperature projections differ slightly among various emission scenarios for the climatology but are largely different for the trend.

[Meta-analysis on efficacy and safety of combination therapy of Aconitum and Western medicine in treatment of rheumatoid arthritis].

To analyze the efficacy and safety of the combination therapy of Aconitum and Western medicine in the treatment of rheumatoid arthritis (RA) by Meta-analysis, and provide evidence for its clinic application for RA. The random clinical trials (RCTs) regarding the combination therapy for treating RA were retrieved in the database of China National Knowledge Infrastructure database, China Science and Technology Journal database, WanFang, Chinese Biomedical Medical Database, PubMed, and Cochrane Library up to July 2017. According to the given inclusion criteria, 8 RCTs involving 659 participants were included, and the included RCTs could be further divided into three subgroups according to the herb type, which were Aconiti Radix (Chuanwu) subgroup (6RCTs), Aconiti Kusnezoffii Radix (Caowu) subgroup (1RCT), and Chuanwu-Caowu subgroup (1RCT). The Meta-analysis results indicated that as compared with Western medicine, the combined use of Aconitum and Western medicine, no matter Chuanwu, Caowu or Chuanwu-Caowu subgroups, could improve the total effective rate of RA (6RCTs, RR=1.19, 95%CI [1.10, 1.28], P<0.000 01), (1 RCT, RR=1.43, 95%CI [1.18, 1.73], P=0.000 2), (1 RCT, RR=1.27, 95%CI [1.02, 1.58], P=0.03) respectively. The combined use of Aconitum and Western medicine was also effective on the number of joint swelling, duration of morning stiffness, patients' handgrip, and the erythrocyte sedimentation rate, C-reactive protein and rheumatoid factor. However, its action was not significant on joint tenderness. And also, in the included RCTs, there were 34 cases of adverse drug reactions/events (ADR/ADE) in the Chuanwu subgroup, while 86 cases in the Western medicine control group. The ADR/ADE incidence was even more lower in Chuanwu-Caowu subgroup, but no difference between Caowu subgroup and Western medicine group. Based on the included RCTs, the combined use of Aconitum and Western medicine could achieve more satisfying efficacy and lower ADRs incidence for RA as compared with Western medicine alone. However, due to the limitation in the not-high quality of included RCTs and the lack of large-scale multi-center research, the results still need to be further validated in the clinic application.