Search

Federated learning (FL) is an attractive paradigm for privacy-aware distributed machine learning, which enables clients to collaboratively learn a global model without sharing clients' data. Recently, many strategies have been proposed to improve the generality of the global model and thus improve FL effect. However, existing strategies either ignore the fairness among clients or sacrifice performance for fairness. They cannot ensure that the gap among clients is as small as possible without sacrificing federated performance. To address this issue, we propose ParetoFed, a new local information aggregation method dedicated to obtaining better federated performance with smaller gap among clients. Specifically, we propose to use multi-objective hyperparameter optimization (HPO) algorithm to gain global models that are both fair and effective. Then, we send Pareto Optimal global models to each client, allowing them to choose the most suitable one as the base to optimize their local model. ParetoFed not only make the global models more fair but also make the selection of local models more personalized, which can further improve the federated performance. Extensive experiments show that ParetoFed outperforms existing FL methods in terms of fairness, and even achieves better federated performance, which demonstrates the significance of our method. [ABSTRACT FROM AUTHOR]

Abstract.Spearman’s footrule is a widely used tool to discover the relationship between two continuous random variables, mainly because of its robustness, natural interpretation, and simplicity of calculation. In this article, we propose a new test statistic based on the sum of squares of pairwise Spearman’s footrule, to test the mutual independence of a high dimensional random vector. The large sample properties of the proposed test method are established. Simulation results show that the new test has higher power than existing test methods under some circumstances. An example with real life data is discussed to illustrate the usefulness and effectiveness of the proposed test statistic. [ABSTRACT FROM AUTHOR]

AbstractThere are lots of methods for complete independence test for high-dimensional random vector. However, it is difficult for practitioners to choose a powerful test because the true alternative hypothesis is unknown. Combining the L2-type with the L∞-type test statistic, we propose a rank-based test method. From a technical point of view, the proposed test is distribution-free and consequently the corresponding critical values or p-values can be obtained by Monte Carlo methods. Compared with permutation or bootstrap test methods, the proposed test method saves calculation cost. Simulation results show that the resulting method has excellent performance with finite sample size. We also provide a real data application to demonstrate the practicality and effectiveness of the proposed test method. [ABSTRACT FROM AUTHOR]

Background Long-term care residents were among the most vulnerable during the COVID-19 pandemic. We estimated vaccine effectiveness of mRNA COVID-19 vaccines in Medicare nursing home residents aged ≥65 years during pre-Delta and high Delta periods. Methods We conducted a retrospective cohort study from 13 December 2020 to 20 November 2021 using Medicare claims data. Exposures included 2 and 3 doses of Pfizer-BioNTech and Moderna COVID-19 vaccines. We used inverse probability weighting and Cox proportional hazards models to estimate absolute and relative vaccine effectiveness. Results Two-dose vaccine effectiveness against COVID-19–related death was 69.8% (95% CI, 65.9%‒73.3%) during the pre-Delta period and 55.7% (49.5%‒61.1%) during the high Delta period, without adjusting for time since vaccination. We observed substantial waning of effectiveness from 65.1% (54.2%‒73.5%) within 6 months from second-dose vaccination to 45.2% (30.6%‒56.7%) ≥6 months after second-dose vaccination in the high Delta period. Three doses provided 88.7% (73.5%‒95.2%) vaccine effectiveness against death, and the incremental benefit of 3 vs 2 doses was 74.6% (40.4%‒89.2%) during high Delta. Among beneficiaries with a prior COVID-19 infection, 3-dose vaccine effectiveness for preventing death was 78.6% (50.0%‒90.8%), and the additional protection of 3 vs 2 doses was 70.0% (30.1%‒87.1%) during high Delta. Vaccine effectiveness estimates against less severe outcomes (eg, infection) were lower. Conclusions This nationwide real-world study demonstrated that mRNA COVID-19 vaccines provided substantial protection against COVID-19–related death. Two-dose protection waned after 6 months. Third doses during the high Delta period provided significant additional protection for individuals with or without a prior COVID-19 infection. [ABSTRACT FROM AUTHOR]