Injuries caused by trauma and neurodegenerative diseases can damage the peripheral nervous system and cause functional deficits.Unlike in the central nervous system,damaged axons in peripheral nerves can be induced to...Injuries caused by trauma and neurodegenerative diseases can damage the peripheral nervous system and cause functional deficits.Unlike in the central nervous system,damaged axons in peripheral nerves can be induced to regenerate in response to intrinsic cues after reprogramming or in a growth-promoting microenvironment created by Schwann cells.However,axon regeneration and repair do not automatically result in the restoration of function,which is the ultimate therapeutic goal but also a major clinical challenge.Transforming growth factor(TGF)is a multifunctional cytokine that regulates various biological processes including tissue repair,embryo development,and cell growth and differentiation.There is accumulating evidence that TGF-βfamily proteins participate in peripheral nerve repair through various factors and signaling pathways by regulating the growth and transformation of Schwann cells;recruiting specific immune cells;controlling the permeability of the blood-nerve barrier,thereby stimulating axon growth;and inhibiting remyelination of regenerated axons.TGF-βhas been applied to the treatment of peripheral nerve injury in animal models.In this context,we review the functions of TGF-βin peripheral nerve regeneration and potential clinical applications.展开更多
针对多变量时序(Multivariate Time Series,MTS)分类中长序列数据难以捕捉时序特征的问题,提出一种基于双向稀疏Transformer的时序分类模型BST(Bidirectional Sparse Transformer),提高了MTS分类任务的准确度.BST模型使用Transformer框...针对多变量时序(Multivariate Time Series,MTS)分类中长序列数据难以捕捉时序特征的问题,提出一种基于双向稀疏Transformer的时序分类模型BST(Bidirectional Sparse Transformer),提高了MTS分类任务的准确度.BST模型使用Transformer框架,构建了一种基于活跃度得分的双向稀疏注意力机制.基于KL散度构建活跃度评价函数,并将评价函数的非对称问题转变为对称权重问题.据此,对原有查询矩阵、键值矩阵进行双向稀疏化,从而降低原Transformer模型中自注意力机制运算的时间复杂度.实验结果显示,BST模型在9个长序列数据集上取得最高平均排名,在临界差异图中领先第2名35.7%,对于具有强时序性的乙醇浓度数据集(Ethanol Concentration,EC),分类准确率提高30.9%.展开更多
针对主流Transformer网络仅对输入像素块做自注意力计算而忽略了不同像素块间的信息交互,以及输入尺度单一导致局部特征细节模糊的问题,本文提出一种基于Transformer并用于处理视觉任务的主干网络ConvFormer. ConvFormer通过所设计的多...针对主流Transformer网络仅对输入像素块做自注意力计算而忽略了不同像素块间的信息交互,以及输入尺度单一导致局部特征细节模糊的问题,本文提出一种基于Transformer并用于处理视觉任务的主干网络ConvFormer. ConvFormer通过所设计的多尺度混洗自注意力模块(Channel-Shuffle and Multi-Scale attention,CSMS)和动态相对位置编码模块(Dynamic Relative Position Coding,DRPC)来聚合多尺度像素块间的语义信息,并在前馈网络中引入深度卷积提高网络的局部建模能力.在公开数据集ImageNet-1K,COCO 2017和ADE20K上分别进行图像分类、目标检测和语义分割实验,ConvFormer-Tiny与不同视觉任务中同量级最优网络RetNetY-4G,Swin-Tiny和ResNet50对比,精度分别提高0.3%,1.4%和0.5%.展开更多
基金supported by the National Natural Science Foundation of China,Nos.31971277 and 31950410551(both to DY)。
文摘Injuries caused by trauma and neurodegenerative diseases can damage the peripheral nervous system and cause functional deficits.Unlike in the central nervous system,damaged axons in peripheral nerves can be induced to regenerate in response to intrinsic cues after reprogramming or in a growth-promoting microenvironment created by Schwann cells.However,axon regeneration and repair do not automatically result in the restoration of function,which is the ultimate therapeutic goal but also a major clinical challenge.Transforming growth factor(TGF)is a multifunctional cytokine that regulates various biological processes including tissue repair,embryo development,and cell growth and differentiation.There is accumulating evidence that TGF-βfamily proteins participate in peripheral nerve repair through various factors and signaling pathways by regulating the growth and transformation of Schwann cells;recruiting specific immune cells;controlling the permeability of the blood-nerve barrier,thereby stimulating axon growth;and inhibiting remyelination of regenerated axons.TGF-βhas been applied to the treatment of peripheral nerve injury in animal models.In this context,we review the functions of TGF-βin peripheral nerve regeneration and potential clinical applications.
文摘针对主流Transformer网络仅对输入像素块做自注意力计算而忽略了不同像素块间的信息交互,以及输入尺度单一导致局部特征细节模糊的问题,本文提出一种基于Transformer并用于处理视觉任务的主干网络ConvFormer. ConvFormer通过所设计的多尺度混洗自注意力模块(Channel-Shuffle and Multi-Scale attention,CSMS)和动态相对位置编码模块(Dynamic Relative Position Coding,DRPC)来聚合多尺度像素块间的语义信息,并在前馈网络中引入深度卷积提高网络的局部建模能力.在公开数据集ImageNet-1K,COCO 2017和ADE20K上分别进行图像分类、目标检测和语义分割实验,ConvFormer-Tiny与不同视觉任务中同量级最优网络RetNetY-4G,Swin-Tiny和ResNet50对比,精度分别提高0.3%,1.4%和0.5%.