There has been many methods in constructing neural network （NN） ensembles, where the method of simultaneous training has succeed in generalization performance and efficiency. But just like regular methods of constru...There has been many methods in constructing neural network （NN） ensembles, where the method of simultaneous training has succeed in generalization performance and efficiency. But just like regular methods of constructing NN ensembles, it follows the two steps, first training component networks, and then combining them. As the two steps being independent, an assumption is used to facilitate interactions among NNs during the training stage. This paper presents a compact ensemble method which integrates the two steps of ensemble construction into one step by attempting to train individual NNs in an ensemble and weigh the individual members adaptively according to their individual performance in the same learning process. This provides an opportunity for the individual NNs to interact with each other based on their real contributions to the ensemble. The classification performance of NN compact ensemble （NNCE） was validated through some benchmark problems in machine learning, including Australian credit card assessment, pima Indians diabetes, heart disease, breast cancer and glass. Compared with other ensembles, the classification error rate of NNCE can be decreased by 0.45% to 68%. In addition, the NNCE was applied to fault diagnosis for rolling element bearing. The 11 time-domain statistical features are extracted as the properties of data, and the NNCE is employed to classify the data. With the results of several experiments, the compact ensemble method is shown to give good generalization performance. The compact ensemble method can recognize the different fault types and various fault degrees of the same fault type.展开更多
Background:Inhibitors of B-cell CLL/Iymphoma 2(Bcl-2)family proteins have shown hope as antitumor drugs.While the notion that it is efficient to coordinate,balance,and neutralize both arms of the anti-apoptotic Bcl-2 ...Background:Inhibitors of B-cell CLL/Iymphoma 2(Bcl-2)family proteins have shown hope as antitumor drugs.While the notion that it is efficient to coordinate,balance,and neutralize both arms of the anti-apoptotic Bcl-2 family has been validated in many cancer cells,the weights of the two arms contributing to apoptosis inhibition have not been explored.This study analyzed the best combination ratio for different Bcl-2 selective inhibitors.Methods:We used a previously established mathematical model to study the weights of Bcl-2(representing both Bcl-2 and BcI-xL in this study)and myeloid cell leukemia-1(Mcl-1).Correlation and single-parameter sensitivity analysis were used to find the major molecular determinants for Bcl-2 and Mcl-1 dependency,as well as their weights.Biological experiments were used to verify the mathematical model.Results:Bcl-2 protein level and Mcl-1 protein level,production,and degradation rates were the major molecular determinants for Bcl-2 and Mcl-1 dependency.The model gained agreement with the experimental assays for ABT-737/A-1210477 and ABT-737/compound 5 combination effect in MCF-7 and MDA-MB-231.Two sets of equations composed of Bcl-2 and Mcl-1 levels were obtained to predict the best combination ratio for Bcl-2 inhibitors with Mcl-1 inhibitors that stabilize and downregulate Mcl-1,respectively.Conclusions:The two sets of equations can be used as tools to bypass time-consuming and laborious experimental screening to predict the best drug combination ratio for treatment.展开更多
基金supported by National Natural Science Foundation of China(Grant No.50575179)National Hi-tech Research and Development Program of China(863 Program,Grant No.2006AA04Z420)
文摘There has been many methods in constructing neural network （NN） ensembles, where the method of simultaneous training has succeed in generalization performance and efficiency. But just like regular methods of constructing NN ensembles, it follows the two steps, first training component networks, and then combining them. As the two steps being independent, an assumption is used to facilitate interactions among NNs during the training stage. This paper presents a compact ensemble method which integrates the two steps of ensemble construction into one step by attempting to train individual NNs in an ensemble and weigh the individual members adaptively according to their individual performance in the same learning process. This provides an opportunity for the individual NNs to interact with each other based on their real contributions to the ensemble. The classification performance of NN compact ensemble （NNCE） was validated through some benchmark problems in machine learning, including Australian credit card assessment, pima Indians diabetes, heart disease, breast cancer and glass. Compared with other ensembles, the classification error rate of NNCE can be decreased by 0.45% to 68%. In addition, the NNCE was applied to fault diagnosis for rolling element bearing. The 11 time-domain statistical features are extracted as the properties of data, and the NNCE is employed to classify the data. With the results of several experiments, the compact ensemble method is shown to give good generalization performance. The compact ensemble method can recognize the different fault types and various fault degrees of the same fault type.
基金This research was supported by the National Natural Science Foundation of China(81430083,81903462 and 82073703)the China Postdoctoral Science Foundation(2018M641694)the Fundamental Research Funds for the Central Universities(DUT20LK28 and DUT20YG133).
文摘Background:Inhibitors of B-cell CLL/Iymphoma 2(Bcl-2)family proteins have shown hope as antitumor drugs.While the notion that it is efficient to coordinate,balance,and neutralize both arms of the anti-apoptotic Bcl-2 family has been validated in many cancer cells,the weights of the two arms contributing to apoptosis inhibition have not been explored.This study analyzed the best combination ratio for different Bcl-2 selective inhibitors.Methods:We used a previously established mathematical model to study the weights of Bcl-2(representing both Bcl-2 and BcI-xL in this study)and myeloid cell leukemia-1(Mcl-1).Correlation and single-parameter sensitivity analysis were used to find the major molecular determinants for Bcl-2 and Mcl-1 dependency,as well as their weights.Biological experiments were used to verify the mathematical model.Results:Bcl-2 protein level and Mcl-1 protein level,production,and degradation rates were the major molecular determinants for Bcl-2 and Mcl-1 dependency.The model gained agreement with the experimental assays for ABT-737/A-1210477 and ABT-737/compound 5 combination effect in MCF-7 and MDA-MB-231.Two sets of equations composed of Bcl-2 and Mcl-1 levels were obtained to predict the best combination ratio for Bcl-2 inhibitors with Mcl-1 inhibitors that stabilize and downregulate Mcl-1,respectively.Conclusions:The two sets of equations can be used as tools to bypass time-consuming and laborious experimental screening to predict the best drug combination ratio for treatment.
