Conventional model transfer techniques,requiring the labelled source data,are not applicable in the privacy-protected medical fields.For the challenging scenarios,recent source data-free domain adaptation(SFDA)has bec...Conventional model transfer techniques,requiring the labelled source data,are not applicable in the privacy-protected medical fields.For the challenging scenarios,recent source data-free domain adaptation(SFDA)has become a mainstream solution but losing focus on the inter-sample class information.This paper proposes a new Credible Local Context Representation approach for SFDA.Our main idea is to exploit the credible local context for more discriminative representation.Specifically,we enhance the source model's discrimination by information regulating.To capture the context,a discovery method is developed that performs fixed steps walking in deep space and takes the credible features in this path as the context.In the epoch-wise adaptation,deep clustering-like training is conducted with two major updates.First,the context for all target data is constructed and then the context-fused pseudo-labels providing semantic guidance are generated.Second,for each target data,a weighting fusion on its context forms the anchored neighbourhood structure;thus,the deep clustering is switched from individual-based to coarse-grained.Also,a new regularisation building is developed on the anchored neighbourhood to drive the deep coarse-grained learning.Experiments on three benchmarks indicate that the proposed method can achieve stateof-the-art results.展开更多
Spatio-temporal variability and dynamics in Sahelian agro-pastoral zones make each local situation a special case. These specificities must be considered to guide the dissemination of agricultural options with a view ...Spatio-temporal variability and dynamics in Sahelian agro-pastoral zones make each local situation a special case. These specificities must be considered to guide the dissemination of agricultural options with a view to sustainable development. The territorial scale of municipalities is not sufficient for this necessary contextualization;the scale of the “village terroir” seems to be a better option. This is the hypothesis we put forward in the framework of the Global Collaboration for Resilient Food Systems program (CRFS), i.e. local context is spatially defined by village terroir. The study is based on data collected through participatory mapping and surveys in “village terroirs” in three regions of Niger (Maradi, Dosso and Tillabéri). Then the links between farm managers and their cultivated land, as well as the spatio-temporal dynamics of local context are analyzed. This study provides evidence of the existence and functional usefulness of the village terroir for farmers, their land management and their activities. It demonstrates the usefulness of contextualizing agricultural options at this scale. Their analysis elucidates the links between “terroirs village” and the specific functioning of the agrosocio-ecosystems acting on each of them, thus laying the systemic and geographical foundations for a model of the spatio- temporal dynamics of “village terroirs”. This initial work has opened up new perspectives in modeling and sustainable development.展开更多
This paper aims to examine New Economics of Labor Migration (NELM) in the northwestern Guangxi, China and investigate the relationships among rural-urban migration, rural household income and local geographical contex...This paper aims to examine New Economics of Labor Migration (NELM) in the northwestern Guangxi, China and investigate the relationships among rural-urban migration, rural household income and local geographical contexts. Stratified sampling and typical case study were adopted and 236 questionnaires were collected from four vil- lages, Daxin, Lixin, Longhe and Yongchang. We analyzed the rural-urban migration rate, household income and local geographical factors, focusing on the ratio of remittance income to total household income. Data descriptions and sta- tistical methods, such as Pearson Chi-square test, Contingency coefficient, Eta, Pearson correlation coefficient, t-test, multiple comparisons (LSD test, Tamhane T2, Dunnett T3 and Dunnet C test) were used. The results are as follows. Rural households’ income is diversified in survey villages so the motivation of rural-urban migration in the study area can be partly explained by NELM. The migration rate of households (the percentage of households with migrants in survey households) in survey villages varies from 50% to 86%, while the proportion of remittance income to house- hold income is in the range of 30% to 80%. In the village of Yongchang, with the least average arable land area per household, the remittance income plays a vital role in household income (80%). And the statistical findings show that the proportion is significantly and negatively correlated with arable land area per household. The conclusion is that di- rect effect of migration, i.e., the contribution of remittance to household income, is negatively correlated with the con- tribution of resources to local income.展开更多
A local and global context representation learning model for Chinese characters is designed and a Chinese word segmentation method based on character representations is proposed in this paper. First, the proposed Chin...A local and global context representation learning model for Chinese characters is designed and a Chinese word segmentation method based on character representations is proposed in this paper. First, the proposed Chinese character learning model uses the semanties of loeal context and global context to learn the representation of Chinese characters. Then, Chinese word segmentation model is built by a neural network, while the segmentation model is trained with the eharaeter representations as its input features. Finally, experimental results show that Chinese charaeter representations can effectively learn the semantic information. Characters with similar semantics cluster together in the visualize space. Moreover, the proposed Chinese word segmentation model also achieves a pretty good improvement on precision, recall and f-measure.展开更多
目的针对远距离红外飞机目标检测中存在的由于成像面积小、辐射强度较弱造成无法充分提取目标特征进而影响检测性能的问题,提出一种基于全局—局部上下文自适应加权融合(adaptive weighted fusion of globallocal context,AWFGLC)机制...目的针对远距离红外飞机目标检测中存在的由于成像面积小、辐射强度较弱造成无法充分提取目标特征进而影响检测性能的问题,提出一种基于全局—局部上下文自适应加权融合(adaptive weighted fusion of globallocal context,AWFGLC)机制的红外飞机目标检测算法。方法基于全局—局部上下文自适应加权融合机制,沿着通道维度随机进行划分与重组,将输入特征图切分为两个特征图。一个特征图使用自注意力进行全局上下文建模,建立目标特征与背景特征之间的相关性,突出目标较显著的特征,使得检测算法更好地感知目标的全局特征。对另一特征图进行窗口划分并在每个窗口内进行最大池化和平均池化以突出目标局部特征,随后使用自注意力对池化特征图进行局部上下文建模,建立目标与其周围邻域的相关性,进一步增强目标特征较弱的部分,使得检测算法更好地感知目标的局部特征。根据目标特点,利用可学习参数的自适应加权融合策略将全局上下文和局部上下文特征图进行聚合,得到包含较完整目标信息的特征图,增强检测算法对目标与背景的判别能力。结果将全局—局部上下文自适应加权融合机制引入YOLOv7(you only look once version 7)并对红外飞机目标进行检测,实验结果表明,提出算法在自制和公开红外飞机数据集的mAP50(mean average precision 50)分别达到97.8%、88.7%,mAP50:95分别达到65.7%、61.2%。结论本文所提出的红外飞机检测算法,优于经典的目标检测算法,能够有效实现红外飞机目标检测。展开更多
基金National Key R&D Program of China,Grant/Award Numbers:2018YFE0203900,2020YFB1313600German Research Foundation,Hamburg Landesforschungsförderungsprojekt Cross,Grant/Award Number:Sonderforschungsbereich Transregio 169+2 种基金Shanghai Artificial Intelligence Innovation Development Special Support Project,Grant/Award Number:3920365001Horizon2020 RISE project STEP2DYNA,Grant/Award Number:691154National Natural Science Foundation of China,Grant/Award Numbers:61773083,62206168,62276048,U1813202。
文摘Conventional model transfer techniques,requiring the labelled source data,are not applicable in the privacy-protected medical fields.For the challenging scenarios,recent source data-free domain adaptation(SFDA)has become a mainstream solution but losing focus on the inter-sample class information.This paper proposes a new Credible Local Context Representation approach for SFDA.Our main idea is to exploit the credible local context for more discriminative representation.Specifically,we enhance the source model's discrimination by information regulating.To capture the context,a discovery method is developed that performs fixed steps walking in deep space and takes the credible features in this path as the context.In the epoch-wise adaptation,deep clustering-like training is conducted with two major updates.First,the context for all target data is constructed and then the context-fused pseudo-labels providing semantic guidance are generated.Second,for each target data,a weighting fusion on its context forms the anchored neighbourhood structure;thus,the deep clustering is switched from individual-based to coarse-grained.Also,a new regularisation building is developed on the anchored neighbourhood to drive the deep coarse-grained learning.Experiments on three benchmarks indicate that the proposed method can achieve stateof-the-art results.
文摘Spatio-temporal variability and dynamics in Sahelian agro-pastoral zones make each local situation a special case. These specificities must be considered to guide the dissemination of agricultural options with a view to sustainable development. The territorial scale of municipalities is not sufficient for this necessary contextualization;the scale of the “village terroir” seems to be a better option. This is the hypothesis we put forward in the framework of the Global Collaboration for Resilient Food Systems program (CRFS), i.e. local context is spatially defined by village terroir. The study is based on data collected through participatory mapping and surveys in “village terroirs” in three regions of Niger (Maradi, Dosso and Tillabéri). Then the links between farm managers and their cultivated land, as well as the spatio-temporal dynamics of local context are analyzed. This study provides evidence of the existence and functional usefulness of the village terroir for farmers, their land management and their activities. It demonstrates the usefulness of contextualizing agricultural options at this scale. Their analysis elucidates the links between “terroirs village” and the specific functioning of the agrosocio-ecosystems acting on each of them, thus laying the systemic and geographical foundations for a model of the spatio- temporal dynamics of “village terroirs”. This initial work has opened up new perspectives in modeling and sustainable development.
基金Under the auspices of the Key Project of National Natural Science Foundation of China (No. 40635029)"985" Proje- ct of Central University for Nationalities (No. 985-2-103-1)
文摘This paper aims to examine New Economics of Labor Migration (NELM) in the northwestern Guangxi, China and investigate the relationships among rural-urban migration, rural household income and local geographical contexts. Stratified sampling and typical case study were adopted and 236 questionnaires were collected from four vil- lages, Daxin, Lixin, Longhe and Yongchang. We analyzed the rural-urban migration rate, household income and local geographical factors, focusing on the ratio of remittance income to total household income. Data descriptions and sta- tistical methods, such as Pearson Chi-square test, Contingency coefficient, Eta, Pearson correlation coefficient, t-test, multiple comparisons (LSD test, Tamhane T2, Dunnett T3 and Dunnet C test) were used. The results are as follows. Rural households’ income is diversified in survey villages so the motivation of rural-urban migration in the study area can be partly explained by NELM. The migration rate of households (the percentage of households with migrants in survey households) in survey villages varies from 50% to 86%, while the proportion of remittance income to house- hold income is in the range of 30% to 80%. In the village of Yongchang, with the least average arable land area per household, the remittance income plays a vital role in household income (80%). And the statistical findings show that the proportion is significantly and negatively correlated with arable land area per household. The conclusion is that di- rect effect of migration, i.e., the contribution of remittance to household income, is negatively correlated with the con- tribution of resources to local income.
基金Supported by the National Natural Science Foundation of China(No.61303179,U1135005,61175020)
文摘A local and global context representation learning model for Chinese characters is designed and a Chinese word segmentation method based on character representations is proposed in this paper. First, the proposed Chinese character learning model uses the semanties of loeal context and global context to learn the representation of Chinese characters. Then, Chinese word segmentation model is built by a neural network, while the segmentation model is trained with the eharaeter representations as its input features. Finally, experimental results show that Chinese charaeter representations can effectively learn the semantic information. Characters with similar semantics cluster together in the visualize space. Moreover, the proposed Chinese word segmentation model also achieves a pretty good improvement on precision, recall and f-measure.
文摘目的针对远距离红外飞机目标检测中存在的由于成像面积小、辐射强度较弱造成无法充分提取目标特征进而影响检测性能的问题,提出一种基于全局—局部上下文自适应加权融合(adaptive weighted fusion of globallocal context,AWFGLC)机制的红外飞机目标检测算法。方法基于全局—局部上下文自适应加权融合机制,沿着通道维度随机进行划分与重组,将输入特征图切分为两个特征图。一个特征图使用自注意力进行全局上下文建模,建立目标特征与背景特征之间的相关性,突出目标较显著的特征,使得检测算法更好地感知目标的全局特征。对另一特征图进行窗口划分并在每个窗口内进行最大池化和平均池化以突出目标局部特征,随后使用自注意力对池化特征图进行局部上下文建模,建立目标与其周围邻域的相关性,进一步增强目标特征较弱的部分,使得检测算法更好地感知目标的局部特征。根据目标特点,利用可学习参数的自适应加权融合策略将全局上下文和局部上下文特征图进行聚合,得到包含较完整目标信息的特征图,增强检测算法对目标与背景的判别能力。结果将全局—局部上下文自适应加权融合机制引入YOLOv7(you only look once version 7)并对红外飞机目标进行检测,实验结果表明,提出算法在自制和公开红外飞机数据集的mAP50(mean average precision 50)分别达到97.8%、88.7%,mAP50:95分别达到65.7%、61.2%。结论本文所提出的红外飞机检测算法,优于经典的目标检测算法,能够有效实现红外飞机目标检测。
