在片上网络(Network on Chip,NoC)系统中,如何完成应用特征图到结构特征图的映射是影响系统实际性能的关键步骤之一。针对NoC系统越发庞大,映射算法耗时也随之增加的问题,提出了自适应模拟退火(Self-Adaptive Simulated Annealing,SASA)...在片上网络(Network on Chip,NoC)系统中,如何完成应用特征图到结构特征图的映射是影响系统实际性能的关键步骤之一。针对NoC系统越发庞大,映射算法耗时也随之增加的问题,提出了自适应模拟退火(Self-Adaptive Simulated Annealing,SASA)的NoC映射算法。采用相对平滑方式实现温度下降过程,针对模拟退火算法易陷于局部最优的缺点,采用自适应方法改变新解生成方式,提高了算法收敛于全局最优的概率。实验结果表明,该算法与常见NoC映射算法(如基于遗传的映射算法)相比,平均性能提升了5.3%,耗时缩短了11.1%。展开更多
In this paper, a second-order implicit-explicit upwind algorithm has been developed for three-dimensional Parabolized Navier-Stokes(PNS) equations. The agreement between the results of the new upwind algorithm and tho...In this paper, a second-order implicit-explicit upwind algorithm has been developed for three-dimensional Parabolized Navier-Stokes(PNS) equations. The agreement between the results of the new upwind algorithm and those of the im- plicit upwind algorithm and its ability in marching a long distance along the stream- wise direction have been shown for the supersonic viscous flow past a sphere-cone body. The CPU time is greatly reduced.展开更多
文摘针对人工跛行检测不够及时,难以发现突发中、重度跛行及轻度跛行行为的问题,该文提出了一种基于正态分布背景统计模型(normal background statistical model,NBSM)与局部循环中心补偿跟踪模型(local circulation center compensation track,LCCCT)和线性斜率最近邻分类(distilling data of KNN,DSKNN)技术的奶牛跛行检测方法。首先利用NBSM模型对奶牛序列图像中的目标奶牛像素区域进行分割,然后对得到的奶牛像素区域利用LCCCT模型提取目标奶牛身体前部像素区域,用其区域通过DSKNN模型提取目标奶牛的头部、颈部以及与颈连接的背部轮廓线拟合直线斜率数据,基于大样本视频序列帧数据将视频集制成轻度跛行、中重度跛行及正常等3类标签的斜率数据集。为了验证算法的有效性,对随机选取的18段奶牛视频进行了验证,其中正常奶牛、轻度跛行奶牛及中重跛行奶牛视频段各6段,获得头部、颈部及背部连接处的拟合直线斜率数据集。在未清洗的数据集上,分别利用SVM、Naive Bayes以及KNN分类算法进行了奶牛跛行的分类检测试验,试验结果表明,SVM与Naive Bayes跛行分类检测正确率均为82.78%,KNN奶牛跛行检测正确率为81.67%。将未清洗的数据集进行清洗后,3类算法的结果表明,KNN分类算法的跛行检测正确率达93.89%,高于SVM分类算法的91.11%及Naive Bayes分类算法的86.11%。上述结果表明通过头部、颈部及背部连接处的拟合直线斜率特性可以正确检测奶牛跛行,未清洗的数据经数据清洗后,KNN分类算法可以取得更好的检测结果。该研究结果对于奶牛跛行疾病的预防、诊断具有重要意义。
文摘在片上网络(Network on Chip,NoC)系统中,如何完成应用特征图到结构特征图的映射是影响系统实际性能的关键步骤之一。针对NoC系统越发庞大,映射算法耗时也随之增加的问题,提出了自适应模拟退火(Self-Adaptive Simulated Annealing,SASA)的NoC映射算法。采用相对平滑方式实现温度下降过程,针对模拟退火算法易陷于局部最优的缺点,采用自适应方法改变新解生成方式,提高了算法收敛于全局最优的概率。实验结果表明,该算法与常见NoC映射算法(如基于遗传的映射算法)相比,平均性能提升了5.3%,耗时缩短了11.1%。
基金The project supported by the National Natural Science Foundation of China
文摘In this paper, a second-order implicit-explicit upwind algorithm has been developed for three-dimensional Parabolized Navier-Stokes(PNS) equations. The agreement between the results of the new upwind algorithm and those of the im- plicit upwind algorithm and its ability in marching a long distance along the stream- wise direction have been shown for the supersonic viscous flow past a sphere-cone body. The CPU time is greatly reduced.
