We design sparse and block feedback gains that minimize the variance amplification (i.e., H <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> norm) of distributed systems. Our approach consists two steps. First, we identify sparsity patterns by incorporating sparsity-promoting penalty functions into optimal control problem, where added terms penalize number communication links in controller. Second, optimize subject to structural...

In this paper, we consider the problem of sensor selection for parameter estimation with correlated measurement noise. We seek optimal activations by formulating an optimization problem, in which error, given trace inverse Bayesian Fisher information matrix, is minimized subject to energy constraints. has been widely used as effective criterion. However, existing information-based methods are limited case uncorrelated noise or weakly due use approximate metrics. By contrast, here derive...

We consider the design of optimal state feedback gains subject to structural constraints on distributed controllers. These are in form sparsity requirements for matrix, implying that each controller has access information from only a limited number subsystems. The minimizer this constrained control problem is sought using augmented Lagrangian method. Notably, approach does not require stabilizing structured gain initialize optimization algorithm. Motivated by structure necessary conditions...

We consider the design of optimal localized feedback gains for one-dimensional formations in which vehicles only use information from their immediate neighbors. The control objective is to enhance coherence formation by making it behave like a rigid lattice. For single-integrator model with symmetric gains, we establish convexity, implying that globally controller can be computed efficiently. also identify class convex problems double-integrators restricting position and uniform diagonal...

We consider a linear quadratic optimal control problem with an additional penalty on the number of communication links in distributed controller. reformulate this combinatorial optimization as sequence weighted ℓ <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</inf> problems, where norm approximates counting links. identify class systems for which can be formulated semidefinite program and therefore its solution computed efficiently....

We are interested in assigning a pre-specified number of nodes as leaders order to minimize the mean-square deviation from consensus stochastically forced networks. This problem arises several applications including control vehicular formations and localization sensor For networks with subject noise, we show that Boolean constraints (a node is either leader or it not) only source nonconvexity. By relaxing these their convex hull obtain lower bound on global optimal value. also use simple but...

We study the optimal design of a conductance network as means for synchronizing given set oscillators. Synchronization is achieved when all oscillator voltages reach consensus, and performance quantified by mean-square deviation from consensus value. formulate optimization problems that address trade-off between synchronization number strength couplings. promote sparsity coupling penalizing interconnection links. For identical oscillators, we establish convexity problem demonstrate can be...

We consider the problem of finding optimal time-periodic sensor schedules for estimating state discrete-time dynamical systems. assume that {multiple} sensors have been deployed and are subject to resource constraints, which limits number times each can be activated over one period periodic schedule. seek an algorithm strikes a balance between estimation accuracy total activations period. make correspondence active nonzero columns estimator gain. formulate optimization in we minimize trace...

In this letter, we study the problem of target tracking based on energy readings sensors. We minimize estimation error by using an extended Kalman filter (EKF). The gain matrix is obtained as solution to optimization in which a sparsity-promoting penalty function added objective. term penalizes number nonzero columns matrix, corresponds active By sparse only few sensors send their measurements fusion center, thereby saving energy. Simulation results show that EKF with can achieve performance...

We consider networks of single-integrator systems, where it is desired to optimally assign a predetermined number systems act as leaders. Performance measured in terms the ℋ <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> norm overall network, and leaders are assumed always follow their state trajectories. demonstrate that, after applying sequence relaxations, problem can be formulated semidefinite program thus solved efficiently....

We study the design of feedback gains that strike a balance between H <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> performance distributed systems and sparsity controller. Our approach consists two steps. First, we identify patterns by incorporating sparsity-promoting penalty functions into problem, where added terms penalize number communication links in Second, optimize subject to structural constraints determined identified...

In the context of distributed estimation, we consider problem sensor collaboration, which refers to act sharing measurements with neighboring sensors prior transmission a fusion center. While incorporating cost aim find optimal sparse collaboration schemes subject certain information or energy constraint. Two types problems are studied: minimum an constraint; and maximum To solve resulting problems, present tractable optimization formulations propose efficient methods that render...

Memristors have recently received significant attention as device-level components for building a novel generation of computing systems. These devices many promising features, such non-volatility, low power consumption, high density, and excellent scalability. The ability to control modify biasing voltages at memristor terminals make them candidates efficiently perform matrix-vector multiplications solve systems linear equations. In this article, we discuss how networks memristors arranged...

We consider the problem of finding optimal feedback gains in presence structural constraints and/or sparsity-promoting penalty functions. Such problems are known to be difficult due their lack convexity. provide an equivalent reformulation optimization such that its source nonconvexity is isolated one nonconvex matrix inequality form Y ≼ X <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">−1</sup> . Furthermore, we preserve gain as variable...

Weight pruning methods of deep neural networks (DNNs) have been demonstrated to achieve a good model rate without loss accuracy, thereby alleviating the significant computation/storage requirements large-scale DNNs. Structured weight proposed overcome limitation irregular network structure and actual GPU acceleration. However, in prior work, (degree sparsity) acceleration are limited (to less than 50%) when accuracy needs be maintained. In this we these limitations by proposing unified,...

We consider the design of optimal static feedback gains for interconnected systems subject to architectural constraints on distributed controller. These are in form sparsity requirements matrix, which means that each controller has access information from only a limited number subsystems. derive necessary conditions optimality structured coupled matrix equations. In general these equations have multiple solutions, is stationary point objective function. For stable open-loop systems, we show...

We examine the leader selection problem in multi-agent dynamical networks where leaders, addition to relative information from their neighbors, also have access own states. are interested selecting an a priori specified number of agents as leaders order minimize total variance stochastically forced network. Combinatorial nature this optimal control makes computation global minimum difficult. propose convex relaxation obtain lower bound on value, and use simple but efficient greedy algorithms...

In this letter, a new sparsity-promoting penalty function is introduced for sensor selection problems in field reconstruction, which has the property of avoiding scenarios where same sensors are successively selected. Using reweighted ℓ <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sub> relaxation xmlns:xlink="http://www.w3.org/1999/xlink">0</sub> norm, problem reformulated as convex quadratic program. order to handle large-scale problems,...

Deep neural networks (DNNs) although achieving human-level performance in many domains, have very large model size that hinders their broader applications on edge computing devices. Extensive research work been conducted DNN compression or pruning. However, most of the previous took heuristic approaches. This proposes a progressive weight pruning approach based ADMM (Alternating Direction Method Multipliers), powerful technique to deal with non-convex optimization problems potentially...

Weight pruning methods of DNNs have been demonstrated to achieve a good model rate without loss accuracy, thereby alleviating the significant computation/storage requirements large-scale DNNs. Structured weight proposed overcome limitation irregular network structure and actual GPU acceleration. However, in prior work (degree sparsity) acceleration are limited (to less than 50%) when accuracy needs be maintained. In this work,we these limitations by proposing unified, systematic framework...

In this paper, we aim to design the optimal sensor collaboration strategy for estimation of time-varying parameters, where refers act sharing measurements with neighboring sensors prior transmission a fusion center. We begin by addressing problem uncorrelated parameters. show that resulting can be transformed into special nonconvex optimization problem, difference convex functions carries all nonconvexity. This specific structure enables use convex-concave procedure obtain near-optimal...

Weight pruning and weight quantization are two important categories of DNN model compression. Prior work on these techniques mainly based heuristics. A recent developed a systematic frame-work using the advanced optimization technique ADMM (Alternating Direction Methods Multipliers), achieving one state-of-art in results. In this work, we first extend such one-shot ADMM-based framework to guarantee solution feasibility provide fast convergence rate, generalize as well. We have further...