With limited number of labeled samples,hyperspectral image(HSI)classification is a difficult Problem in current research.The graph neural network(GNN)has emerged as an approach to semi-supervised classification,and th...With limited number of labeled samples,hyperspectral image(HSI)classification is a difficult Problem in current research.The graph neural network(GNN)has emerged as an approach to semi-supervised classification,and the application of GNN to hyperspectral images has attracted much attention.However,in the existing GNN-based methods a single graph neural network or graph filter is mainly used to extract HSI features,which does not take full advantage of various graph neural networks(graph filters).Moreover,the traditional GNNs have the problem of oversmoothing.To alleviate these shortcomings,we introduce a deep hybrid multi-graph neural network(DHMG),where two different graph filters,i.e.,the spectral filter and the autoregressive moving average(ARMA)filter,are utilized in two branches.The former can well extract the spectral features of the nodes,and the latter has a good suppression effect on graph noise.The network realizes information interaction between the two branches and takes good advantage of different graph filters.In addition,to address the problem of oversmoothing,a dense network is proposed,where the local graph features are preserved.The dense structure satisfies the needs of different classification targets presenting different features.Finally,we introduce a GraphSAGEbased network to refine the graph features produced by the deep hybrid network.Extensive experiments on three public HSI datasets strongly demonstrate that the DHMG dramatically outperforms the state-ofthe-art models.展开更多
The titania hollow microspheres with incontinuous multicavities were successfully fabricated via an oil/ water(O/W) emulsion process accompanied by sol–gel reaction in the presence of polyvinylpyrrolidone(PVP). In th...The titania hollow microspheres with incontinuous multicavities were successfully fabricated via an oil/ water(O/W) emulsion process accompanied by sol–gel reaction in the presence of polyvinylpyrrolidone(PVP). In the one-step route, the addition of PVP to the tetrabutyl titanate(TBT) 1-octanol solution as the oil phase of the O/W emulsion clearly expands the size of the cavities inside the microspheres. The and atoleine alters the polarity of the oil phase to affect the interior structure significantly. The Span 80 is used as a stabilizer to preserve spherical shape. A preliminary mechanism based on phaseseparation for the structural evolution of titania hollow microspheres with multicavities is suggested. Zirconia and alumina hollow microspheres with incontinuous multicavities can also be prepared by this one-step route successfully.展开更多
The light-matter interaction between plasmonic nanocavity and exciton at the sub-diffraction limit is a central research field in nanophotonics.Here,we demonstrated the vertical distribution of the light-matter intera...The light-matter interaction between plasmonic nanocavity and exciton at the sub-diffraction limit is a central research field in nanophotonics.Here,we demonstrated the vertical distribution of the light-matter interactions at~1 nm spatial resolution by coupling A excitons of MoS2 and gap-mode plasmonic nanocavities.Moreover,we observed the significant photoluminescence(PL)enhancement factor reaching up to 2800 times,which is attributed to the Purcell effect and large local density of states in gap-mode plasmonic nanocavities.Meanwhile,the theoretical calculations are well reproduced and support the experimental results.展开更多
文摘With limited number of labeled samples,hyperspectral image(HSI)classification is a difficult Problem in current research.The graph neural network(GNN)has emerged as an approach to semi-supervised classification,and the application of GNN to hyperspectral images has attracted much attention.However,in the existing GNN-based methods a single graph neural network or graph filter is mainly used to extract HSI features,which does not take full advantage of various graph neural networks(graph filters).Moreover,the traditional GNNs have the problem of oversmoothing.To alleviate these shortcomings,we introduce a deep hybrid multi-graph neural network(DHMG),where two different graph filters,i.e.,the spectral filter and the autoregressive moving average(ARMA)filter,are utilized in two branches.The former can well extract the spectral features of the nodes,and the latter has a good suppression effect on graph noise.The network realizes information interaction between the two branches and takes good advantage of different graph filters.In addition,to address the problem of oversmoothing,a dense network is proposed,where the local graph features are preserved.The dense structure satisfies the needs of different classification targets presenting different features.Finally,we introduce a GraphSAGEbased network to refine the graph features produced by the deep hybrid network.Extensive experiments on three public HSI datasets strongly demonstrate that the DHMG dramatically outperforms the state-ofthe-art models.
基金supported by the National Natural Science Foundation of China (No. 51372225)Zhejiang Provincial Natural Science Foundation of China (No. LY13B010001)
文摘The titania hollow microspheres with incontinuous multicavities were successfully fabricated via an oil/ water(O/W) emulsion process accompanied by sol–gel reaction in the presence of polyvinylpyrrolidone(PVP). In the one-step route, the addition of PVP to the tetrabutyl titanate(TBT) 1-octanol solution as the oil phase of the O/W emulsion clearly expands the size of the cavities inside the microspheres. The and atoleine alters the polarity of the oil phase to affect the interior structure significantly. The Span 80 is used as a stabilizer to preserve spherical shape. A preliminary mechanism based on phaseseparation for the structural evolution of titania hollow microspheres with multicavities is suggested. Zirconia and alumina hollow microspheres with incontinuous multicavities can also be prepared by this one-step route successfully.
基金supported by the National Key Research and Development Program of China(2019YFA0705400,2020YFB1505800,2019YFD0901100.and 2021YFA12015021.the National Natural Science Foundation of China(21925404,22021001,22002128,21991151,and 92161118).the Science and Technology Planning Project of Fujian Province(2021Y0104).the State Key Laboratory of Fine Chemicals Dalian University of Technology(KF2002 and the“111”Project(B17027).
文摘The light-matter interaction between plasmonic nanocavity and exciton at the sub-diffraction limit is a central research field in nanophotonics.Here,we demonstrated the vertical distribution of the light-matter interactions at~1 nm spatial resolution by coupling A excitons of MoS2 and gap-mode plasmonic nanocavities.Moreover,we observed the significant photoluminescence(PL)enhancement factor reaching up to 2800 times,which is attributed to the Purcell effect and large local density of states in gap-mode plasmonic nanocavities.Meanwhile,the theoretical calculations are well reproduced and support the experimental results.
