We consider the questions connected with the approximation of a real continuous 1-periodic functions and give a new proof of the equivalence of the special Boman-Shapiro modulus of continuity with Peetre’s K-function...We consider the questions connected with the approximation of a real continuous 1-periodic functions and give a new proof of the equivalence of the special Boman-Shapiro modulus of continuity with Peetre’s K-functional. We also prove Jackson’s inequality for the approximation by trigonometric polynomials.展开更多
An inverse theorem for the best weighted polynomial approximation of a function in L<sub>w<sub>α</sub></sub><sup>p</sup>（S） is established. We also investigate Besov spaces gener...An inverse theorem for the best weighted polynomial approximation of a function in L<sub>w<sub>α</sub></sub><sup>p</sup>（S） is established. We also investigate Besov spaces generated by Freud weight and their connection with algebraic polynomial approximation in L<sub>p</sub>（R）w<sub>λ</sub>, where w<sub>α</sub> is a Jacobi-type weight on S, 0【p≤∞, S is a simplex and W<sub>λ</sub> is a Freud weight. For Ditzian-Totik K-functionals on L<sub>P</sub>（S）, 1≤P≤∞, we obtain a new equivalence expression.展开更多
针对视频异常事件的时空相关性学习对检测性能存在重要影响的问题,提出了基于融合双支特征的带约束损失的视频异常检测方法(Dual-branch Feature Fusion Based Constrained Multi-loss Video Anomaly Detection,DBF-CML-transMIL)。该...针对视频异常事件的时空相关性学习对检测性能存在重要影响的问题,提出了基于融合双支特征的带约束损失的视频异常检测方法(Dual-branch Feature Fusion Based Constrained Multi-loss Video Anomaly Detection,DBF-CML-transMIL)。该方法考虑多示例学习中片段的显著性和相关性,利用多层线性神经网络学习各片段的空间显著性特征,并设计级联Transformer融合模块来学习示例间的多层时序相关性;然后利用多损失模型对融合特征进行多loss监督学习,以丰富预测的多样性;针对现有top-k的离散性问题,提出了带约束机制的滑窗top-k强化异常事件的相关性。在ShanghaiTech和UCF-Crime数据集上的对比实验与消融实验表明,DBF-CML-transMIL的异常检测曲线下面积(Area Under Curve,AUC)分别达到97.33%和83.82%;各模块都能有效提升视频异常事件检测的性能。展开更多
We investigate Besov spaces and their connection with trigonometric polynomial approximation in L_p[－π,π], algebraic polynomial approximation in L_p[-1,1], algebraic polynomial approximation in L_p(S), and entir...We investigate Besov spaces and their connection with trigonometric polynomial approximation in L_p[－π,π], algebraic polynomial approximation in L_p[-1,1], algebraic polynomial approximation in L_p(S), and entire function of exponential type approximation in Lp(R), and characterize K-functionals for certain pairs of function spaces including (Lp [-π,π], B_s~α (Lp[-π，π])), (L_p(R),B_s~α (Lp(R))), (Lp[-1,1],B_s~α (Lp[-1,1])), and (Lp(S),B_s~α (Lp(S))), where 0<s<, 0<p<1, S is a simple polytope and 0<α<r.展开更多
自平衡机器人是验证各种控制算法的经典装置,研究传输时滞对其控制系统的影响具有重要意义。基于李雅普诺夫-克拉索夫斯基(Lyapunov-Krasovskii,L-K)泛函方法讨论了自平衡机器人的控制问题。首先,建立了直流电机的线性化模型,并应用拉...自平衡机器人是验证各种控制算法的经典装置,研究传输时滞对其控制系统的影响具有重要意义。基于李雅普诺夫-克拉索夫斯基(Lyapunov-Krasovskii,L-K)泛函方法讨论了自平衡机器人的控制问题。首先,建立了直流电机的线性化模型,并应用拉格朗日方程法建立了自平衡机器人的线性数学模型;然后,进一步考虑传输时滞环节,建立基于多闭环比例积分微分(proportional integral differential,PID)控制器的自平衡机器人控制系统的整体状态空间模型;最后,应用广义自由矩阵积分不等式,建立了低保守性的L-K稳定性判据,在此基础上通过MATLAB中的线性矩阵不等式(linear matrix inequality,LMI)工具箱去求解PID控制增益对时滞稳定裕度的影响。仿真结果表明,所提出的系统稳定性判据不仅有效,而且具有较低的保守性。展开更多
文摘We consider the questions connected with the approximation of a real continuous 1-periodic functions and give a new proof of the equivalence of the special Boman-Shapiro modulus of continuity with Peetre’s K-functional. We also prove Jackson’s inequality for the approximation by trigonometric polynomials.
文摘An inverse theorem for the best weighted polynomial approximation of a function in L<sub>w<sub>α</sub></sub><sup>p</sup>（S） is established. We also investigate Besov spaces generated by Freud weight and their connection with algebraic polynomial approximation in L<sub>p</sub>（R）w<sub>λ</sub>, where w<sub>α</sub> is a Jacobi-type weight on S, 0【p≤∞, S is a simplex and W<sub>λ</sub> is a Freud weight. For Ditzian-Totik K-functionals on L<sub>P</sub>（S）, 1≤P≤∞, we obtain a new equivalence expression.
文摘针对视频异常事件的时空相关性学习对检测性能存在重要影响的问题,提出了基于融合双支特征的带约束损失的视频异常检测方法(Dual-branch Feature Fusion Based Constrained Multi-loss Video Anomaly Detection,DBF-CML-transMIL)。该方法考虑多示例学习中片段的显著性和相关性,利用多层线性神经网络学习各片段的空间显著性特征,并设计级联Transformer融合模块来学习示例间的多层时序相关性;然后利用多损失模型对融合特征进行多loss监督学习,以丰富预测的多样性;针对现有top-k的离散性问题,提出了带约束机制的滑窗top-k强化异常事件的相关性。在ShanghaiTech和UCF-Crime数据集上的对比实验与消融实验表明,DBF-CML-transMIL的异常检测曲线下面积(Area Under Curve,AUC)分别达到97.33%和83.82%;各模块都能有效提升视频异常事件检测的性能。
基金This project is supported by the National Natural Science Foundation of China.
文摘We investigate Besov spaces and their connection with trigonometric polynomial approximation in L_p[－π,π], algebraic polynomial approximation in L_p[-1,1], algebraic polynomial approximation in L_p(S), and entire function of exponential type approximation in Lp(R), and characterize K-functionals for certain pairs of function spaces including (Lp [-π,π], B_s~α (Lp[-π，π])), (L_p(R),B_s~α (Lp(R))), (Lp[-1,1],B_s~α (Lp[-1,1])), and (Lp(S),B_s~α (Lp(S))), where 0<s<, 0<p<1, S is a simple polytope and 0<α<r.
文摘自平衡机器人是验证各种控制算法的经典装置,研究传输时滞对其控制系统的影响具有重要意义。基于李雅普诺夫-克拉索夫斯基(Lyapunov-Krasovskii,L-K)泛函方法讨论了自平衡机器人的控制问题。首先,建立了直流电机的线性化模型,并应用拉格朗日方程法建立了自平衡机器人的线性数学模型;然后,进一步考虑传输时滞环节,建立基于多闭环比例积分微分(proportional integral differential,PID)控制器的自平衡机器人控制系统的整体状态空间模型;最后,应用广义自由矩阵积分不等式,建立了低保守性的L-K稳定性判据,在此基础上通过MATLAB中的线性矩阵不等式(linear matrix inequality,LMI)工具箱去求解PID控制增益对时滞稳定裕度的影响。仿真结果表明,所提出的系统稳定性判据不仅有效,而且具有较低的保守性。
