Recovering shape and motion by a dynamic system for low-rank matrix approximation in L 1 norm

To recover motion and shape matrices from a matrix of tracking feature points on a rigid object under orthography, we can do low-rank matrix approximation of the tracking matrix with its each column minus the row mean vector of the matrix. To obtain the row mean vector, usually 4-rank matrix approximation is used to recover the missing entries. Then, 3-rank matrix approximation is used to recover the shape and motion. Obviously, the procedure is not convenient. In this paper, we build a cost function which calculates the shape matrix, motion matrix as well as the row mean vector at the same time. The function is in L ₁ norm, and is not smooth everywhere. To optimize the function, a continuous-time dynamic system is newly proposed. With time going on, the product of the shape and rotation matrices becomes closer and closer, in L ₁-norm sense, to the tracking matrix with each its column minus the mean vector. A parameter is implanted into the system for improving the calculating efficiency, and the influence of the parameter on approximation accuracy and computational efficiency are theoretically studied and experimentally confirmed. The experimental results on a large number of synthetic data and a real application of structure from motion demonstrate the effectiveness and efficiency of the proposed method. The proposed system is also applicable to general low-rank matrix approximation in L ₁ norm, and this is also experimentally demonstrated.     

Face recognition by sparse discriminant analysis via joint L2,1-norm minimization

Recently, joint feature selection and subspace learning, which can perform feature selection and subspace learning simultaneously, is proposed and has encouraging ability on face recognition. In the literature, a framework of utilizing L_2,1-norm penalty term has also been presented, but some important algorithms cannot be covered, such as Fisher Linear Discriminant Analysis and Sparse Discriminant Analysis. Therefore, in this paper, we add L_2,1-norm penalty term on FLDA and propose a feasible solution by transforming its nonlinear model into linear regression type. In addition, we modify the optimization model of SDA by replacing elastic net with L_2,1-norm penalty term and present its optimization method. Experiments on three standard face databases illustrate FLDA and SDA via L_2,1-norm penalty term can significantly improve their recognition performance, and obtain inspiring results with low computation cost and for low-dimension feature.     

Fuzzy measures for a correlation coefficient of fuzzy numbers under TW(the weakest t-norm)-based fuzzy arithmetic operations

Recently, the sup-min convolution based on Zadeh’s extension principle has been used by Liu and Kao [Fuzzy measures for correlation coefficient of fuzzy numbers, Fuzzy Sets and Systems 128 (2002) 267-275], to calculate a fuzzy correlation coefficient. They used a mathematical programming approach to derive fuzzy measures based on the classical definition of the correlation coefficient. It is well known that T_W (the weakest t-norm)-based addition and multiplication preserve the shape of L-R fuzzy numbers. In this paper, we consider the computational aspect of the T_W-based extension principle when the principle is applied to a correlation coefficient of L-R fuzzy numbers. We give the exact solution of a fuzzy correlation coefficient without programming or the aid of computer resources.     

Perturbing Bézier coefficients for best constrained degree reduction in the L2-norm

This paper first shows how the Bézier coefficients of a given degree n polynomial are perturbed so that it can be reduced to a degree m (<n) polynomial with the constraint that continuity of a prescribed order is preserved at the two endpoints. The perturbation vector, which consists of the perturbation coefficients, is determined by minimizing a weighted Euclidean norm. The optimal degree n−1 approximation polynomial is explicitly given in Bézier form. Next the paper proves that the problem of finding a best L₂-approximation over the interval [0,1] for constrained degree reduction is equivalent to that of finding a minimum perturbation vector in a certain weighted Euclidean norm. The relevant weights are derived. This result is applied to computing the optimal constrained degree reduction of parametric Bézier curves in the L₂-norm.     

Extensive theoretical/numerical comparative studies on H2 and generalised H2 norms in sampled-data systems

This paper is concerned with linear time-invariant (LTI) sampled-data systems (by which we mean sampled-data systems with LTI generalised plants and LTI controllers) and studies their H₂ norms from the viewpoint of impulse responses and generalised H₂ norms from the viewpoint of the induced norms from L₂ to L_∞. A new definition of the H₂ norm of LTI sampled-data systems is first introduced through a sort of intermediate standpoint of those for the existing two definitions. We then establish unified treatment of the three definitions of the H₂ norm through a matrix function G(τ) defined on the sampling interval [0, h). This paper next considers the generalised H₂ norms, in which two types of the L_∞ norm of the output are considered as the temporal supremum magnitude under the spatial 2-norm and ∞-norm of a vector-valued function. We further give unified treatment of the generalised H₂ norms through another matrix function F(θ) which is also defined on [0, h). Through a close connection between G(τ) and F(θ), some theoretical relationships between the H₂ and generalised H₂ norms are provided. Furthermore, appropriate extensions associated with the treatment of G(τ) and F(θ) to the closed interval [0, h] are discussed to facilitate numerical computations and comparisons of the H₂ and generalised H₂ norms. Through theoretical and numerical studies, it is shown that the two generalised H₂ norms coincide with neither of the three H₂ norms of LTI sampled-data systems even though all the five definitions coincide with each other when single-output continuous-time LTI systems are considered as a special class of LTI sampled-data systems. To summarise, this paper clarifies that the five control performance measures are mutually related with each other but they are also intrinsically different from each other.     

Synchronization of multiple chaotic FitzHugh-Nagumo neurons with gap junctions under external electrical stimulation

This paper discusses the synchronization of three coupled chaotic FitzHugh-Nagumo (FHN) neurons with different gap junctions under external electrical stimulation. A nonlinear control law that guarantees the asymptotic synchronization of coupled neurons (with reduced computations) is proposed. The developed control law incorporates the synchronization error between two slave neurons in addition to the conventionally considered synchronization errors between the master and the slave neurons, which make the proposed scheme computationally more efficient. Further, a novel L₂ gain reduction criterion has been developed for multi-input multi-output systems with non-zero initial conditions, and is applied to robust synchronization of FHN neurons under L₂ norm bounded disturbance and uncertainties. Furthermore, a robust adaptive nonlinear control law is developed, which is capable of handling variations in nonlinear part of synchronization error dynamics, without using any neural-network-based training-oriented adaptive scheme. The proposed control schemes ensure global synchronization with computational simplicity, easy way of design and implementation and avoiding extra measurements. The results obtained with the proposed control laws are verified through numerical simulations.     

Reduced-order controllers for the H∞ control problem with unstable invariant zeros

This paper addresses the existence and design methods of reduced-order controllers for the H_∞ control problem with unstable invariant zeros in the state-space realization of the transfer function matrix from the control input to the controlled error or from the exogenous input to the observation output, where the realization is induced from a stabilizable and detectable realization of the generalized plant. This paper presents a new controller degree bound for the H_∞ control problem in terms of the minimal rank of the system matrix pencils of these two transfer function matrices in the unstable region. When the unstable invariant zero exists, this paper shows that reduced-order controllers with orders strictly less than that of the generalized plant exist if the H_∞ control problem is solvable. Moreover, this paper shows that the computational problem of finding the controllers with the new degree bound is convex by providing two linear matrix inequality-based design methods (algorithms) for constructing the reduced-order controllers. The results developed in this paper are valid both for the continuous- and discrete-time H_∞ control problems.     

基于L2范数最小化的实时目标跟踪

在贝叶斯推理框架下,基于稀疏表示的跟踪算法能够较好地处理目标在视频场景中的各种复杂的外观变化,取得较为鲁棒的跟踪效果,但算法的计算复杂度很高,很难满足实时性要求。针对稀疏跟踪算法的这一问题,提出了一种基于l2范数最小化的实时目标跟踪算法。将PCA子空间目标表示与l2范数最小化进行结合,去除稀疏跟踪算法中常用的琐碎模板集,建立了基于l2范数最小化的目标表示模型以及将遮挡等因素考虑在内的观测似然度函数。在大量的实验测试集上的对比实验结果显示,该算法和多个非常优秀的跟踪算法相比,可以达到相同甚至更高的跟踪精度,而且在多个测试集上可以达到每秒20帧的速度。该算法可以很好地应对视频监控场景中遮挡、光线突变、尺度变化和非刚性形变等干扰,同时算法复杂度低,满足了实时要求。     

The H2-control for jump linear systems: cluster observations of the Markov state

The H₂-norm control problem of discrete-time Markov jump linear systems is addressed in this paper when part of, or the total of the Markov states is not accessible to the controller. The non-observed part of the Markov states is grouped in a number of clusters of observations; the case with a single cluster retrieves the situation when no Markov state is observed. The control action is provided in linear feedback form, which is invariant on each cluster, and this restricted complexity setting is adopted, aiming at computable solutions. We explore a recent result by de Oliveira, Bernussou, and Geromel (Systems Control Lett. 37 (1999) 261) involving an LMI characterization to establish a H₂ solution that is stabilizing in the mean square sense. The novelty of the method is that it can handle in LMI form the situation ranging from no Markov state observation to complete state observation. In addition, when the state observation is complete, the optimal H₂-norm solution is retrieved.     

H2 and H∞ norm computations of linear continuous-time periodic systems via the skew analysis of frequency response operators

The H₂ and H_∞ norm computations of finite-dimensional linear continuous-time periodic (FDLCP) systems through the frequency response operators defined by steady-state analysis are discussed. By the skew truncation, the H₂ norm can be reached to any degree of accuracy by that of an asymptotically equivalent linear time-invariant (LTI) continuous-time system. The H_∞ norm can be approximated by the maximum singular value of the frequency response of an asymptotically equivalent LTI continuous-time system over a certain frequency range via the modified skew truncation. By the latter result, a Hamiltonian test is proved for FDLCP systems in an LTI fashion, based on which a modified bisection algorithm is developed.     

四旋翼无人机L1 自适应块控反步姿态控制器设计

针对四旋翼无人机的姿态控制问题,提出一种L1自适应块控反步控制方法.将四旋翼姿态运动模型转化为一类多输入多输出不确定非线性系统的形式;根据该系统严格反馈的结构特点,对外回路设计了块控反步控制器;针对内回路存在的外部干扰和内部参数摄动等不确定性,引入L1自适应控制思想补偿其影响.稳定性分析表明闭环系统内所有信号一致有界.仿真和姿态稳定实验验证了所提控制策略的有效性和鲁棒性.     

An approximation ratio for biclustering

The problem of biclustering consists of the simultaneous clustering of rows and columns of a matrix such that each of the submatrices induced by a pair of row and column clusters is as uniform as possible. In this paper we approximate the optimal biclustering by applying one-way clustering algorithms independently on the rows and on the columns of the input matrix. We show that such a solution yields a worst-case approximation ratio of under L₁-norm for 0-1 valued matrices, and of 2 under L₂-norm for real valued matrices.     

A linear-time algorithm for linearL1 approximation of points

In this paper we present a linear-time algorithm for approximating a set ofn points by a linear function, or a line, that minimizes theL ₁ norm. The algorithmic complexity of this problem appears not to have been investigated, although anO(n ³) naive algorithm can be easily obtained based on some simple characteristics of an optimumL ₁ solution. Our linear-time algorithm is optimal within a constant factor and enables us to use linearL ₁ approximation of many points in practice. The complexity ofL ₁ linear approximation of a piecewise linear function is also touched upon.     

Convergence and convergence rate of the balanced realization truncations for infinite-dimensional discrete-time systems

In this paper, we study approximation via balanced realizations for a large class of infinite-dimensional discrete-time linear systems. We give properties of the truncated system and prove that the approximation is L₂-convergent. In the case of nuclear systems, we prove convergence in nuclear norm and give an estimation of the L_∞-convergence rate.     

Robust switching-type H∞ filter design for linear uncertain systems with time-varying delay

This paper is concerned with the problem of robust H_∞ filter design for a class of linear uncertain systems with time-varying delay. The uncertainty parameters are supposed to be time-varying, unknown, but bounded, which appear affinely in the matrices of the considered system model. Based on linear matrix inequality (LMI) method and switching laws, a new switching-type filter is designed to guarantee the asymptotic stability and H_∞ performance level of the filtering error systems. The key feature is that the new proposed filter parameters are switching between certain fixed gains automatically via the designed switching law. It is shown that the new filter design method is less conservative than the traditional fixed gain filter design method. An example is given to illustrate the validity of the proposed design.     

A parametric optimization approach to H∞ and H2 strong stabilization

This paper presents an approach for designing stable MIMO H_∞ and H₂ controllers by directly computing the norm-constrained stable transfer matrices Q in the H_∞ and H₂ suboptimal controller parameterizations. This is done by first converting the H₂ and H_∞ strong stabilization problems into some nonlinear unconstrained optimization problems through explicit parameterization of the norm-constrained Q's for any fixed order. Then, a two-stage numerical search is carried out by using a combination of a genetic algorithm and a quasi-Newton algorithm in order to reach an optimal solution. The effectiveness of the proposed algorithms is illustrated through some benchmark numerical examples.     

半张量积压缩感知模型的l0-范数解

目的 半张量积压缩感知模型是一种可以有效降低压缩感知过程中随机观测矩阵所占存储空间的新方法,利用该模型可以成倍降低观测矩阵所需的存储空间。为寻求基于该模型新的重构方法,同时提升降维后观测矩阵的重构性能,提出一种采用光滑高斯函数拟合l₀-范数方法进行重构。方法 构建降维随机观测矩阵,对原始信号进行采样;构建可微且期望值为零的光滑高斯函数来拟合不连续的l₀-范数,采用最速下降法进行重构,最终得到稀疏信号的估计值。结果 实验分别采用1维稀疏信号和2维图像信号进行测试,并从重构概率、收敛速度、重构信号的峰值信噪比等角度进行了测试和比较。验证结果表明,本文所述算法的重构概率、收敛速度较该模型的l_q-范数（0 <q <1）方法有一定的提升,且当观测矩阵大小降低为通常的1/64,甚至1/256时,仍能保持较高的重构性能。结论 本文所述的重构算法,能在更大程度上降低观测矩阵的大小,同时基本保持重构的精度。     

MIMO离散时间系统混合l1/H2控制

研究了MIMO(多输入多输出)离散时间系统的混合l₁/H₂优化问题,该问题可描述为最优化一个传递函数矩阵的l₁范数同时保证另一个传递函数矩阵的H₂范数满足预定的指标.研究了最优目标函数值关于H₂范数指标的连续性.证明了MIMO系统混合l₁/H₂控制问题最优解的存在性.由于基于标定-Q(scaled-Q)方法求解MIMO混合l₁/H₂问题,避免了进行零点插值运算的困难.通过求解有限维非线性规划问题可得到最优目标值的收敛的上下界.     

Optimal time-weighted H 2 model reduction for Markovian jump systems

This article addresses the optimal time-weighted H ₂ model reduction problem for Markovian jump linear systems. That is, for a given mean square stable Markovian jump system, our aim is to find a mean square stable jump system of lower order such that the time-weighted H ₂ norm of the corresponding error system is minimised. The time-weighted H ₂ norm of the system is first defined, and then a computational method is constructed. The computation requires the solution of two sets of recursive Lyapunov-type linear matrix equations associated with the Markovian jump system. To solve the optimal time-weighted H ₂ model reduction problem, we propose a gradient flow method for its solution. A necessary condition for minimality is derived, and a computational procedure is provided to obtain the minimising reduced-order model. The necessary condition generalises the standard result for systems when Markov jumps and the time-weighting term do not appear. Finally, two numerical examples are given to demonstrate the effectiveness of the proposed approach.     

2 guaranteed cost control for uncertain continuous-time linear systems

This paper proposes a solution to the ₂ guaranteed cost control problem for uncertain, continuous-time linear systems. It consists of the determination of a constant state feedback stabilizing matrix gain and a ₂-norm upper bound, valid for all feasible models. The uncertainties are only assumed to be convex-bounded, a concept which generalizes the important case of interval matrices uncertainties. The results follow from a new parameterization of all stabilizing gains over a convex set. As an additional property, the above mentioned ₂-norm upper bound reduces to the minimum ₂ cost in case of precisely known linear systems.