With the popularity of online payment, how to perform creditcard fraud detection more accurately has also become a hot issue. And withthe emergence of the adaptive boosting algorithm (Adaboost), credit cardfraud detec...With the popularity of online payment, how to perform creditcard fraud detection more accurately has also become a hot issue. And withthe emergence of the adaptive boosting algorithm (Adaboost), credit cardfraud detection has started to use this method in large numbers, but thetraditional Adaboost is prone to overfitting in the presence of noisy samples.Therefore, in order to alleviate this phenomenon, this paper proposes a newidea: using the number of consecutive sample misclassifications to determinethe noisy samples, while constructing a penalty factor to reconstruct thesample weight assignment. Firstly, the theoretical analysis shows that thetraditional Adaboost method is overfitting in a noisy training set, which leadsto the degradation of classification accuracy. To this end, the penalty factorconstructed by the number of consecutive misclassifications of samples isused to reconstruct the sample weight assignment to prevent the classifierfrom over-focusing on noisy samples, and its reasonableness is demonstrated.Then, by comparing the penalty strength of the three different penalty factorsproposed in this paper, a more reasonable penalty factor is selected.Meanwhile, in order to make the constructed model more in line with theactual requirements on training time consumption, the Adaboost algorithmwith adaptive weight trimming (AWTAdaboost) is used in this paper, so thepenalty factor-based AWTAdaboost (PF_AWTAdaboost) is finally obtained.Finally, PF_AWTAdaboost is experimentally validated against other traditionalmachine learning algorithms on credit card fraud datasets and otherdatasets. The results show that the PF_AWTAdaboost method has betterperformance, including detection accuracy, model recall and robustness, thanother methods on the credit card fraud dataset. And the PF_AWTAdaboostmethod also shows excellent generalization performance on other datasets.From the experimental results, it is shown that the PF_AWTAdaboost algorithmhas better classification performance.展开更多
Realizing high performance in both n-type and p-type materials is essential for designing efficient ther-moelectric devices.However,the doping bottleneck is often encountered,i.e.,only one type of conduction can be re...Realizing high performance in both n-type and p-type materials is essential for designing efficient ther-moelectric devices.However,the doping bottleneck is often encountered,i.e.,only one type of conduction can be realized.As one example,p-type CdSb with high thermoelectric performance has been discovered for several decades,while its n-type counterpart has rarely been reported.In this work,the calculated band structure of CdSb demonstrates that the valley degeneracy is as large as ten for the conduction band,and it is only two for the valence band.Therefore,the n-type CdSb can potentially realize an ex-ceptional thermoelectric performance.Experimentally,the n-type conduction has been successfully real-ized by tuning the stoichiometry of CdSb.By further doping indium at the Cd site,an improved room-temperature electron concentration has been achieved.Band modeling predicts an optimal electron con-centration of∼2.0×1019 cm−3,which is higher than the current experimental values.Therefore,future optimization of the n-type CdSb should mainly focus on identifying practical approaches to optimize the electron concentration.展开更多
Abdominal aortic aneurysm(AAA)and atherosclerosis(AS)have considerable similarities in clinical risk factors and molecular pathogenesis.The aim of our study was to investigate the differences between AAA and AS from t...Abdominal aortic aneurysm(AAA)and atherosclerosis(AS)have considerable similarities in clinical risk factors and molecular pathogenesis.The aim of our study was to investigate the differences between AAA and AS from the perspective of metabolomics,and to explore the potential mechanisms of differential metabolites via integration analysis with transcriptomics.Plasma samples from 32 AAA and 32 AS patients were applied to characterize the metabolite profiles using untargeted liquid chromatography-mass spectrometry(LC-MS).A total of 18 remarkably different metabolites were identified,and a combination of seven metabolites could potentially serve as a biomarker to distinguish AAA and AS,with an area under the curve(AUC)of0.93.Subsequently,we analyzed both the metabolomics and transcriptomics data and found that seven metabolites,especially 2’-deoxy-D-ribose(2 d DR),were significantly correlated with differentially expressed genes.In conclusion,our study presents a comprehensive landscape of plasma metabolites in AAA and AS patients,and provides a research direction for pathogenetic mechanisms in atherosclerotic AAA.展开更多
基金This research was funded by Innovation and Entrepreneurship Training Program for College Students in Hunan Province in 2022(3915).
文摘With the popularity of online payment, how to perform creditcard fraud detection more accurately has also become a hot issue. And withthe emergence of the adaptive boosting algorithm (Adaboost), credit cardfraud detection has started to use this method in large numbers, but thetraditional Adaboost is prone to overfitting in the presence of noisy samples.Therefore, in order to alleviate this phenomenon, this paper proposes a newidea: using the number of consecutive sample misclassifications to determinethe noisy samples, while constructing a penalty factor to reconstruct thesample weight assignment. Firstly, the theoretical analysis shows that thetraditional Adaboost method is overfitting in a noisy training set, which leadsto the degradation of classification accuracy. To this end, the penalty factorconstructed by the number of consecutive misclassifications of samples isused to reconstruct the sample weight assignment to prevent the classifierfrom over-focusing on noisy samples, and its reasonableness is demonstrated.Then, by comparing the penalty strength of the three different penalty factorsproposed in this paper, a more reasonable penalty factor is selected.Meanwhile, in order to make the constructed model more in line with theactual requirements on training time consumption, the Adaboost algorithmwith adaptive weight trimming (AWTAdaboost) is used in this paper, so thepenalty factor-based AWTAdaboost (PF_AWTAdaboost) is finally obtained.Finally, PF_AWTAdaboost is experimentally validated against other traditionalmachine learning algorithms on credit card fraud datasets and otherdatasets. The results show that the PF_AWTAdaboost method has betterperformance, including detection accuracy, model recall and robustness, thanother methods on the credit card fraud dataset. And the PF_AWTAdaboostmethod also shows excellent generalization performance on other datasets.From the experimental results, it is shown that the PF_AWTAdaboost algorithmhas better classification performance.
基金supported by the Shenzhen Science and Tech-nology Program (No.KQTD20200820113045081)the State Key Laboratory of Advanced Welding and Joining,Harbin Institute of Technology.J.M.acknowledges the financial support from the National Natural Science Foundation of China (No.52101248)+6 种基金Shenzhen fundamental research projects (No.JCYJ20210324132808020)the start-up funding of Shenzhen,and the start-up funding of Harbin Institute of Technology (Shenzhen).Q.Z.acknowledges the financial support from the National Nat-ural Science Foundation of China (Nos.52172194 and 51971081)the Natural Science Foundation for Distinguished Young Scholars of Guangdong Province of China (No.2020B1515020023)the Natural Science Foundation for Distinguished Young Scholars of Shenzhen (No.RCJC20210609103733073)the Key Project of Shenzhen Fundamental Research Projects (No.JCYJ20200109113418655)F.C.acknowledges the financial support from the National Natural Science Foundation of China (No.51871081)H.L.acknowledges the financial support from the National Natural Science Foundation of China (No.62174044).
文摘Realizing high performance in both n-type and p-type materials is essential for designing efficient ther-moelectric devices.However,the doping bottleneck is often encountered,i.e.,only one type of conduction can be realized.As one example,p-type CdSb with high thermoelectric performance has been discovered for several decades,while its n-type counterpart has rarely been reported.In this work,the calculated band structure of CdSb demonstrates that the valley degeneracy is as large as ten for the conduction band,and it is only two for the valence band.Therefore,the n-type CdSb can potentially realize an ex-ceptional thermoelectric performance.Experimentally,the n-type conduction has been successfully real-ized by tuning the stoichiometry of CdSb.By further doping indium at the Cd site,an improved room-temperature electron concentration has been achieved.Band modeling predicts an optimal electron con-centration of∼2.0×1019 cm−3,which is higher than the current experimental values.Therefore,future optimization of the n-type CdSb should mainly focus on identifying practical approaches to optimize the electron concentration.
基金supported by the National Natural Science Foundation of China(Nos.51890894,81770481,and 82070492)the Chinese Academy of Medical SciencesInnovation Fund for Medical Sciences(CIFMS 2017-I2M-1-008)。
文摘Abdominal aortic aneurysm(AAA)and atherosclerosis(AS)have considerable similarities in clinical risk factors and molecular pathogenesis.The aim of our study was to investigate the differences between AAA and AS from the perspective of metabolomics,and to explore the potential mechanisms of differential metabolites via integration analysis with transcriptomics.Plasma samples from 32 AAA and 32 AS patients were applied to characterize the metabolite profiles using untargeted liquid chromatography-mass spectrometry(LC-MS).A total of 18 remarkably different metabolites were identified,and a combination of seven metabolites could potentially serve as a biomarker to distinguish AAA and AS,with an area under the curve(AUC)of0.93.Subsequently,we analyzed both the metabolomics and transcriptomics data and found that seven metabolites,especially 2’-deoxy-D-ribose(2 d DR),were significantly correlated with differentially expressed genes.In conclusion,our study presents a comprehensive landscape of plasma metabolites in AAA and AS patients,and provides a research direction for pathogenetic mechanisms in atherosclerotic AAA.