Quorum sensing (QS) refers to the cell communication through signaling molecules that regulate many important biological functions of bacteria by monitoring their population density. Although a wide spectrum of studie...Quorum sensing (QS) refers to the cell communication through signaling molecules that regulate many important biological functions of bacteria by monitoring their population density. Although a wide spectrum of studies on the QS system mechanisms have been carried out in experiments, mathematical modeling to explore the QS system has become a powerful approach as well. In this paper, we review the research progress of network modeling in bacterial QS to capture the system's underlying mechanisms. There are four types of QS system models for bacteria: the Gram-negative QS system model, the Gram-positive QS system model, the model for both Gram-negative and Gram-positive QS system, and the synthetic QS system model. These QS system models are mostly described by the ordinary differential equations (ODE) or partial differential equations (PDE) to study the changes of signaling molecule dynamics in time and space and the cell population density variations. Besides the deterministic simulations, the stochastic modeling approaches have also been introduced to discuss the noise effects on kinetics in QS systems. Taken together, these current modeling efforts advance our understanding of the QS system by providing systematic and quantitative dynamics description, which can hardly be obtained in experiments.展开更多
The sandy land of the western part of Jilin Province is located in the ecotone of semi-humid and semi-arid area in the temperate zone of China. The sandy desertification has widely spread in the...The sandy land of the western part of Jilin Province is located in the ecotone of semi-humid and semi-arid area in the temperate zone of China. The sandy desertification has widely spread in the region because of the vulnerable natural conditions and the unreasonable human activity; as a result of this, the precious land resources and the economic development in the area have been seriously impacted. In this paper, the sandy land ecologic environment geographic information system is established based on the multi-spectral, multi-temporal Landsat TM images and field investigation. The comprehensive indexes of sandy desertification extent assessment which include vegetation degradation, wind erosion extent and soil depth are presented to classify the sandy land in western Jilin into three levels--slight, moderate and severe sandy desertification with the support of GIS platform. The results demonstrate that the sandy desertification has been partly controlled in the past twenty years, except some small sites. However, this doesn't necessarily mean that there is nothing for more concern. The two main causes of sandy desertification have not been eliminated yet, one is its natural factor, especially the physical and chemical characters of sandy soil and dry climate; another is the immoderate economic activity of human being that has highly accelerated the sandy desertification process.展开更多
语义分割是遥感影像分析中的重要技术之一。现有方法(如基于深度卷积神经网络的方法等)虽然在语义分割中取得了显著进展,但往往需要大量训练数据。基于图模型的马尔可夫随机场模型(Markov random field model,MRF)提出了一种不依赖训练...语义分割是遥感影像分析中的重要技术之一。现有方法(如基于深度卷积神经网络的方法等)虽然在语义分割中取得了显著进展,但往往需要大量训练数据。基于图模型的马尔可夫随机场模型(Markov random field model,MRF)提出了一种不依赖训练数据的无监督语义分割思路,可以有效地刻画地物空间关系,并对地物空间分布的统计规律进行建模。但现有的MRF模型方法通常建立在基于像素或对象的单一粒度基元上,难以充分利用影像信息,语义分割效果不佳。针对上述问题,引入交替方向乘子法(alternative direction method of multiplier,ADMM)并将其离散化,提出了一种像素与对象基元协同的MRF模型无监督语义分割方法(MRF-ADMM)。首先构建像素基元和对象基元两个概率图,其中像素基元概率图用于刻画影像的细节信息,保持语义分割的边界;对象基元概率图用于描述较大范围的空间关系,以应对遥感影像地物内部的高异质性,使分割结果中地物内部具有良好的区域完整性。在模型求解过程中,针对像素和对象基元的特点,提出了一种离散化的ADMM方法,并将其用于两种基元类别标记的传递与更新,实现像素基元细节信息和对象基元区域信息的协同优化。高分二号和航拍影像等不同数据库不同类型遥感影像的语义分割实验结果表明,相较于现有的MRF模型,提出的MRF-ADMM方法能有效地协同不同粒度基元的优点,优化语义分割结果。展开更多
为更好地利用单词词性包含的语义信息和伴随单词出现时的非自然语言上下文信息,提出动态调整语义的词性加权多模态情感分析(part of speech weighted multi-modal sentiment analysis model with dynamic semantics adjustment,PW-DS)模...为更好地利用单词词性包含的语义信息和伴随单词出现时的非自然语言上下文信息,提出动态调整语义的词性加权多模态情感分析(part of speech weighted multi-modal sentiment analysis model with dynamic semantics adjustment,PW-DS)模型.该模型以自然语言为主体,分别使用基于Transformer的双向编码器表示(bidirectional encoder representation from Transformers,BERT)模型、广义自回归预训练(generalized autoregressive pretraining for language understanding,XLNet)模型和一种鲁棒优化的BERT预训练(robustly optimized BERT pretraining approach,RoBERTa)模型为文本模态做词嵌入编码;创建动态调整语义模块将自然语言和非自然语言信息有效结合;设计词性加权模块,提取单词词性并赋权以优化情感判别.与张量融合网络和低秩多模态融合等当前先进模型的对比实验结果表明,PW-DS模型在公共数据集CMU-MOSI和CMU-MOSEI上的平均绝对误差分别达到了0.607和0.510,二分类准确率分别为89.02%和86.93%,优于对比模型.通过消融实验分析了不同模块对模型效果的影响,验证了模型的有效性.展开更多
基金Project supported by the National Natural Science Foundation of China(Grant Nos.11704318,11675134,and 11874310)the China Postdoctoral Science Foundation(Grant No.2016M602071).
文摘Quorum sensing (QS) refers to the cell communication through signaling molecules that regulate many important biological functions of bacteria by monitoring their population density. Although a wide spectrum of studies on the QS system mechanisms have been carried out in experiments, mathematical modeling to explore the QS system has become a powerful approach as well. In this paper, we review the research progress of network modeling in bacterial QS to capture the system's underlying mechanisms. There are four types of QS system models for bacteria: the Gram-negative QS system model, the Gram-positive QS system model, the model for both Gram-negative and Gram-positive QS system, and the synthetic QS system model. These QS system models are mostly described by the ordinary differential equations (ODE) or partial differential equations (PDE) to study the changes of signaling molecule dynamics in time and space and the cell population density variations. Besides the deterministic simulations, the stochastic modeling approaches have also been introduced to discuss the noise effects on kinetics in QS systems. Taken together, these current modeling efforts advance our understanding of the QS system by providing systematic and quantitative dynamics description, which can hardly be obtained in experiments.
基金Department of Science & Technology of Jilin Province No.1999010
文摘The sandy land of the western part of Jilin Province is located in the ecotone of semi-humid and semi-arid area in the temperate zone of China. The sandy desertification has widely spread in the region because of the vulnerable natural conditions and the unreasonable human activity; as a result of this, the precious land resources and the economic development in the area have been seriously impacted. In this paper, the sandy land ecologic environment geographic information system is established based on the multi-spectral, multi-temporal Landsat TM images and field investigation. The comprehensive indexes of sandy desertification extent assessment which include vegetation degradation, wind erosion extent and soil depth are presented to classify the sandy land in western Jilin into three levels--slight, moderate and severe sandy desertification with the support of GIS platform. The results demonstrate that the sandy desertification has been partly controlled in the past twenty years, except some small sites. However, this doesn't necessarily mean that there is nothing for more concern. The two main causes of sandy desertification have not been eliminated yet, one is its natural factor, especially the physical and chemical characters of sandy soil and dry climate; another is the immoderate economic activity of human being that has highly accelerated the sandy desertification process.
文摘语义分割是遥感影像分析中的重要技术之一。现有方法(如基于深度卷积神经网络的方法等)虽然在语义分割中取得了显著进展,但往往需要大量训练数据。基于图模型的马尔可夫随机场模型(Markov random field model,MRF)提出了一种不依赖训练数据的无监督语义分割思路,可以有效地刻画地物空间关系,并对地物空间分布的统计规律进行建模。但现有的MRF模型方法通常建立在基于像素或对象的单一粒度基元上,难以充分利用影像信息,语义分割效果不佳。针对上述问题,引入交替方向乘子法(alternative direction method of multiplier,ADMM)并将其离散化,提出了一种像素与对象基元协同的MRF模型无监督语义分割方法(MRF-ADMM)。首先构建像素基元和对象基元两个概率图,其中像素基元概率图用于刻画影像的细节信息,保持语义分割的边界;对象基元概率图用于描述较大范围的空间关系,以应对遥感影像地物内部的高异质性,使分割结果中地物内部具有良好的区域完整性。在模型求解过程中,针对像素和对象基元的特点,提出了一种离散化的ADMM方法,并将其用于两种基元类别标记的传递与更新,实现像素基元细节信息和对象基元区域信息的协同优化。高分二号和航拍影像等不同数据库不同类型遥感影像的语义分割实验结果表明,相较于现有的MRF模型,提出的MRF-ADMM方法能有效地协同不同粒度基元的优点,优化语义分割结果。
文摘为更好地利用单词词性包含的语义信息和伴随单词出现时的非自然语言上下文信息,提出动态调整语义的词性加权多模态情感分析(part of speech weighted multi-modal sentiment analysis model with dynamic semantics adjustment,PW-DS)模型.该模型以自然语言为主体,分别使用基于Transformer的双向编码器表示(bidirectional encoder representation from Transformers,BERT)模型、广义自回归预训练(generalized autoregressive pretraining for language understanding,XLNet)模型和一种鲁棒优化的BERT预训练(robustly optimized BERT pretraining approach,RoBERTa)模型为文本模态做词嵌入编码;创建动态调整语义模块将自然语言和非自然语言信息有效结合;设计词性加权模块,提取单词词性并赋权以优化情感判别.与张量融合网络和低秩多模态融合等当前先进模型的对比实验结果表明,PW-DS模型在公共数据集CMU-MOSI和CMU-MOSEI上的平均绝对误差分别达到了0.607和0.510,二分类准确率分别为89.02%和86.93%,优于对比模型.通过消融实验分析了不同模块对模型效果的影响,验证了模型的有效性.
