A reproducing kernel collocation method based on strong formulation is introduced for transient dynamics. To study the stability property of this method, an algorithm based on the von Neumann hypothesis is proposed to...A reproducing kernel collocation method based on strong formulation is introduced for transient dynamics. To study the stability property of this method, an algorithm based on the von Neumann hypothesis is proposed to predict the critical time step. A numerical test is conducted to validate the algorithm. The numerical critical time step and the predicted critical time step are in good agreement. The results are compared with those obtained based on the radial basis collocation method, and they axe in good agreement. Several important conclusions for choosing a proper support size of the reproducing kernel shape function are given to improve the stability condition.展开更多
The transports of the dynamic biochemical signals in the non-reversing pulsatile flows in the mixing microchannel of a Y-shaped microfluidic device are ana- lyzed. The results show that the mixing micro-channel acts a...The transports of the dynamic biochemical signals in the non-reversing pulsatile flows in the mixing microchannel of a Y-shaped microfluidic device are ana- lyzed. The results show that the mixing micro-channel acts as a low-pass filter, and the biochemical signals are nonlinearly modulated by the pulsatile flows, which depend on the biochemical signal frequency, the flow signal frequency, and the biochemical signal transporting distance. It is concluded that, the transfer characteristics of the dynamic biochemical signals, which are transported in the time-varying flows, should be carefully considered for better loading biochemical signals on the cells cultured on the bottom of the microfluidic channel.展开更多
针对全监督视频实例分割网络训练数据高度依赖精细掩码标注,时间和人工成本过高,导致智能机器无法快速适应新场景的问题,提出一种端到端的掩码生成动态调控弱监督视频实例分割(Weakly Supervised Video Instance Segmentation,WSVIS)网...针对全监督视频实例分割网络训练数据高度依赖精细掩码标注,时间和人工成本过高,导致智能机器无法快速适应新场景的问题,提出一种端到端的掩码生成动态调控弱监督视频实例分割(Weakly Supervised Video Instance Segmentation,WSVIS)网络。为克服初始掩码预测层通道维度突降导致的实例激活特征丢失问题,构建多级特征融合模块,利用特征复用策略预测初始实例特征并融合相对位置信息生成初始预测掩码。然后,提出动态调控机制在通道和空间维度上建立掩码特征依赖关系,强化初始预测掩码与实例感知信息之间的动态交互。最后,网络设计二元颜色相似性生成伪亲和标签取代精细掩码标注,联合边界框与掩码一致性损失实现仅边界框标注的弱监督视频实例分割。实验结果表明,在BoxSet和YT-VIS数据集上,WSVIS网络能达到与全监督网络相近的分割精度和分割效果,同时能够满足实时推理要求,为智能机器快速适应新场景实现实时环境感知和理解提供了理论支撑和算法依据。展开更多
基金Project supported by the Western Transport Technical Project of Ministry of Transport of China(No. 2009318000046)
文摘A reproducing kernel collocation method based on strong formulation is introduced for transient dynamics. To study the stability property of this method, an algorithm based on the von Neumann hypothesis is proposed to predict the critical time step. A numerical test is conducted to validate the algorithm. The numerical critical time step and the predicted critical time step are in good agreement. The results are compared with those obtained based on the radial basis collocation method, and they axe in good agreement. Several important conclusions for choosing a proper support size of the reproducing kernel shape function are given to improve the stability condition.
基金Project supported by the National Natural Science Foundation of China(Nos.11172060 and11672065)
文摘The transports of the dynamic biochemical signals in the non-reversing pulsatile flows in the mixing microchannel of a Y-shaped microfluidic device are ana- lyzed. The results show that the mixing micro-channel acts as a low-pass filter, and the biochemical signals are nonlinearly modulated by the pulsatile flows, which depend on the biochemical signal frequency, the flow signal frequency, and the biochemical signal transporting distance. It is concluded that, the transfer characteristics of the dynamic biochemical signals, which are transported in the time-varying flows, should be carefully considered for better loading biochemical signals on the cells cultured on the bottom of the microfluidic channel.
文摘针对全监督视频实例分割网络训练数据高度依赖精细掩码标注,时间和人工成本过高,导致智能机器无法快速适应新场景的问题,提出一种端到端的掩码生成动态调控弱监督视频实例分割(Weakly Supervised Video Instance Segmentation,WSVIS)网络。为克服初始掩码预测层通道维度突降导致的实例激活特征丢失问题,构建多级特征融合模块,利用特征复用策略预测初始实例特征并融合相对位置信息生成初始预测掩码。然后,提出动态调控机制在通道和空间维度上建立掩码特征依赖关系,强化初始预测掩码与实例感知信息之间的动态交互。最后,网络设计二元颜色相似性生成伪亲和标签取代精细掩码标注,联合边界框与掩码一致性损失实现仅边界框标注的弱监督视频实例分割。实验结果表明,在BoxSet和YT-VIS数据集上,WSVIS网络能达到与全监督网络相近的分割精度和分割效果,同时能够满足实时推理要求,为智能机器快速适应新场景实现实时环境感知和理解提供了理论支撑和算法依据。
基金Supported by National Science and Technology Special Program(No.2011ZX04004-041)National Natural Science Foundation of China(No.90923023)Doctoral Program Foundation of Higher Education of China(No.20092302110055)
