This paper studies the generalization capability of feedforward neural networks (FNN).The mechanism of FNNs for classification is investigated from the geometric and probabilistic viewpoints. It is pointed out that th...This paper studies the generalization capability of feedforward neural networks (FNN).The mechanism of FNNs for classification is investigated from the geometric and probabilistic viewpoints. It is pointed out that the outputs of the output layer in the FNNs for classification correspond to the estimates of posteriori probability of the input pattern samples with desired outputs 1 or 0. The theorem for the generalized kernel function in the radial basis function networks (RBFN) is given. For an 2-layer perceptron network (2-LPN). an idea of using extended samples to improve generalization capability is proposed. Finally. the experimental results of radar target classification are given to verify the generaliztion capability of the RBFNs.展开更多
Many studies have shown the influence of protein corona(PC)on the active targeting capability of ligand-modified nanoparticles;however,the influence of clinical status on PC composition and targeting capacity is rarel...Many studies have shown the influence of protein corona(PC)on the active targeting capability of ligand-modified nanoparticles;however,the influence of clinical status on PC composition and targeting capacity is rarely discussed.In this study,when transferrinmodified PEGylated polystyrene nanoparticles(Tf-PNs)is intravenously injected into mice with non-small cell lung cancer(NSCLC)comorbid with type 2 diabetes mellitus(T2DM),more Tf-PNs accumulated in the tumor tissue than in those of NSCLC model mice.This indicated that PC derived fromdifferent states of disease changed the active targeting ability of Tf-PNs.To explain the occurrence of this phenomenon,our analysis of PC from different disease states revealed that Tf(transferrin)modification had no significant effect on the formation of PC,and that the PC from the NSCLC comorbid with T2DM model contained more proteins like fibrin and clusterin.This work demonstrates the impacts of comorbidity,such as with T2DM,on the active targeting capability of ligand-modified nanoparticles,and the results promote the application of nanoparticles for precision medicine.展开更多
文摘This paper studies the generalization capability of feedforward neural networks (FNN).The mechanism of FNNs for classification is investigated from the geometric and probabilistic viewpoints. It is pointed out that the outputs of the output layer in the FNNs for classification correspond to the estimates of posteriori probability of the input pattern samples with desired outputs 1 or 0. The theorem for the generalized kernel function in the radial basis function networks (RBFN) is given. For an 2-layer perceptron network (2-LPN). an idea of using extended samples to improve generalization capability is proposed. Finally. the experimental results of radar target classification are given to verify the generaliztion capability of the RBFNs.
基金supported by the Key Research and Development Program of Science and Technology Department of the Sichuan Province 111 Project(B18035).
文摘Many studies have shown the influence of protein corona(PC)on the active targeting capability of ligand-modified nanoparticles;however,the influence of clinical status on PC composition and targeting capacity is rarely discussed.In this study,when transferrinmodified PEGylated polystyrene nanoparticles(Tf-PNs)is intravenously injected into mice with non-small cell lung cancer(NSCLC)comorbid with type 2 diabetes mellitus(T2DM),more Tf-PNs accumulated in the tumor tissue than in those of NSCLC model mice.This indicated that PC derived fromdifferent states of disease changed the active targeting ability of Tf-PNs.To explain the occurrence of this phenomenon,our analysis of PC from different disease states revealed that Tf(transferrin)modification had no significant effect on the formation of PC,and that the PC from the NSCLC comorbid with T2DM model contained more proteins like fibrin and clusterin.This work demonstrates the impacts of comorbidity,such as with T2DM,on the active targeting capability of ligand-modified nanoparticles,and the results promote the application of nanoparticles for precision medicine.
