The ongoing impact of the novel coronavirus disease 2019(COVID-19)on work and daily life persists as we transition from emergency to normal circumstances.The continuous mutation of viral strains has resulted in a shif...The ongoing impact of the novel coronavirus disease 2019(COVID-19)on work and daily life persists as we transition from emergency to normal circumstances.The continuous mutation of viral strains has resulted in a shift from a single strain to multiple cross-strains,contributing to the spread of the epidemic.Variations in infection rates of the same strain occur because of the implementation of diverse preventive measures at different times.This study investigated the dynamics of the pandemic in the presence of concurrent strains.Building on the classical Susceptible,Exposed,Infected,and Recovered(SEIR)model,a robust piecewise multi-strain cross-epidemic trend prediction model was proposed that employs the Hodges–Lehmann estimator to handle uncertain and contamination-prone epidemic information.A comparative analysis of epidemic spread trend curves across diverse populations using different robust methods revealed the superiority of the Hodges–Lehmann estimator-based model over the traditional method.The accurate prediction results of the model demonstrate its high reliability in tracking the changing trend of the COVID-19 outbreak,thereby supporting its implementation in subsequent epidemic prevention and control measures.展开更多
Rotating machinery is critical to industrial systems,necessitating robust anomaly detection(AD)to ensure operational safety and prevent failures.However,in real-world scenarios,monitoring data is typically unlabeled a...Rotating machinery is critical to industrial systems,necessitating robust anomaly detection(AD)to ensure operational safety and prevent failures.However,in real-world scenarios,monitoring data is typically unlabeled and often consists of normal samples contaminated with a small proportion of unknown anomalies.To address this,this paper proposes a diffusion-based AD method,Anomaly Detection Denoising Diffusion Probabilistic Model(AD-DDPM)for robust AD.The method employs a U-attention-net to capture local and global features and introduces a filtered contrastive mechanism to mitigate the impact of contaminated training data.By leveraging the probabilistic nature of diffusion models,AD-DDPM effectively models normal data distributions,achieving superior AD even with polluted samples.Experimental validation on fault simulation datasets demonstrates the method’s exceptional performance,outperforming traditional machine learning and deep learning baselines.The proposed approach offers a promising solution for reliable health monitoring in industrial settings.展开更多
In this study, the two-sided Empirical Bayes test （EBT） rules for the parameter of continuous one-parameter exponential family with contaminated data （errors in variables） are constructed by a deconvolution kernel...In this study, the two-sided Empirical Bayes test （EBT） rules for the parameter of continuous one-parameter exponential family with contaminated data （errors in variables） are constructed by a deconvolution kernel method. The asymptotically optimal uniformly over a class of prior distributions and uniform rates of convergence, which depends on two types of the error distribu- tions for the proposed EBT rules, are obtained under suitable con- ditions. Finally, an example about the main results of this paper is given.展开更多
In this paper, the following contaminated linear model is considered:y i=(1-ε)x τ iβ+z i, 1≤i≤n,where r.v.'s ｛ y i ｝ are contaminated with errors ｛ z i ｝. To assume that the errors have the fin...In this paper, the following contaminated linear model is considered:y i=(1-ε)x τ iβ+z i, 1≤i≤n,where r.v.'s ｛ y i ｝ are contaminated with errors ｛ z i ｝. To assume that the errors have the finite moment of order 2 only. The non parametric estimation of contaminated coefficient ε and regression parameter β are established, and the strong consistency and convergence rate almost surely of the estimators are obtained. A simulated example is also given to show the visual performance of the estimations.展开更多
基金This work was supported by National Science Foundation of China with Grant No.721040202022 Science and Technology Think Tank Young Talent Program with Grant No.20220615zz07110051。
文摘The ongoing impact of the novel coronavirus disease 2019(COVID-19)on work and daily life persists as we transition from emergency to normal circumstances.The continuous mutation of viral strains has resulted in a shift from a single strain to multiple cross-strains,contributing to the spread of the epidemic.Variations in infection rates of the same strain occur because of the implementation of diverse preventive measures at different times.This study investigated the dynamics of the pandemic in the presence of concurrent strains.Building on the classical Susceptible,Exposed,Infected,and Recovered(SEIR)model,a robust piecewise multi-strain cross-epidemic trend prediction model was proposed that employs the Hodges–Lehmann estimator to handle uncertain and contamination-prone epidemic information.A comparative analysis of epidemic spread trend curves across diverse populations using different robust methods revealed the superiority of the Hodges–Lehmann estimator-based model over the traditional method.The accurate prediction results of the model demonstrate its high reliability in tracking the changing trend of the COVID-19 outbreak,thereby supporting its implementation in subsequent epidemic prevention and control measures.
基金supported by The National Natural Science Foundation of China under Grant(5247512)National Key Lab of Aerospace Power System and Plasma Technology Foundation(APSPT202304002).
文摘Rotating machinery is critical to industrial systems,necessitating robust anomaly detection(AD)to ensure operational safety and prevent failures.However,in real-world scenarios,monitoring data is typically unlabeled and often consists of normal samples contaminated with a small proportion of unknown anomalies.To address this,this paper proposes a diffusion-based AD method,Anomaly Detection Denoising Diffusion Probabilistic Model(AD-DDPM)for robust AD.The method employs a U-attention-net to capture local and global features and introduces a filtered contrastive mechanism to mitigate the impact of contaminated training data.By leveraging the probabilistic nature of diffusion models,AD-DDPM effectively models normal data distributions,achieving superior AD even with polluted samples.Experimental validation on fault simulation datasets demonstrates the method’s exceptional performance,outperforming traditional machine learning and deep learning baselines.The proposed approach offers a promising solution for reliable health monitoring in industrial settings.
基金Supported by the Fundamental Research Funds for the Central Universities of China(2013-Ia-040)
文摘In this study, the two-sided Empirical Bayes test （EBT） rules for the parameter of continuous one-parameter exponential family with contaminated data （errors in variables） are constructed by a deconvolution kernel method. The asymptotically optimal uniformly over a class of prior distributions and uniform rates of convergence, which depends on two types of the error distribu- tions for the proposed EBT rules, are obtained under suitable con- ditions. Finally, an example about the main results of this paper is given.
文摘In this paper, the following contaminated linear model is considered:y i=(1-ε)x τ iβ+z i, 1≤i≤n,where r.v.'s ｛ y i ｝ are contaminated with errors ｛ z i ｝. To assume that the errors have the finite moment of order 2 only. The non parametric estimation of contaminated coefficient ε and regression parameter β are established, and the strong consistency and convergence rate almost surely of the estimators are obtained. A simulated example is also given to show the visual performance of the estimations.
