Random projection algorithm is of interest for constrained optimization when the constraint set is not known in advance or the projection operation on the whole constraint set is computationally prohibitive. This pape...
详细信息
Random projection algorithm is of interest for constrained optimization when the constraint set is not known in advance or the projection operation on the whole constraint set is computationally prohibitive. This paper presents a distributed random projection algorithm for constrained convex optimization problems that can be used by multiple agents connected over a time-varying network, where each agent has its own objective function and its own constrained set. We prove that the iterates of all agents converge to the same point in the optimal set almost surely. Experiments on distributed support vector machines demonstrate good performance of the algorithm.
In this paper, a distributedmulti-agent scheduling system (MASS) based on co-operative approach is proposed to solve static and dynamic job shop scheduling problems (JSSP). The proposed system is composed of two kind...
详细信息
In this paper, a distributedmulti-agent scheduling system (MASS) based on co-operative approach is proposed to solve static and dynamic job shop scheduling problems (JSSP). The proposed system is composed of two kinds of agents, Supervisor agents and Resource agents. The Supervisor agent decomposes JSSP into interrelated sub-problems and the Resource agents co-operate, through a distributed approach of local idle time minimisation, to solve this problem which is known as one of the most difficult NP-hard problems. Computational results are presented to show the efficiency of MASS in static job shop scheduling. Then, a comparison of the computational results between MASS and some common dispatching rules, on dynamic job arrivals, is studied in terms of effectiveness and stability. Finally, the developed system is validated within an illustrative example, to demonstrate the feasibility of MASS.
The goal of decentralized optimization over a network is to optimize a global objective formed by a sum of local (possibly nonsmooth) convex functions using only local computation and communication. It arises in vario...
详细信息
The goal of decentralized optimization over a network is to optimize a global objective formed by a sum of local (possibly nonsmooth) convex functions using only local computation and communication. It arises in various application domains, including distributed tracking and localization, multi-agent coordination, estimation in sensor networks, and large-scale machine learning. We develop and analyze distributed algorithms based on dual subgradient averaging, and we provide sharp bounds on their convergence rates as a function of the network size and topology. Our analysis allows us to clearly separate the convergence of the optimization algorithm itself and the effects of communication dependent on the network structure. We show that the number of iterations required by our algorithm scales inversely in the spectral gap of the network, and confirm this prediction's sharpness both by theoretical lower bounds and simulations for various networks. Our approach includes the cases of deterministic optimization and communication, as well as problems with stochastic optimization and/or communication.
We consider a distributedmulti-agent network system where each agent has its own convex objective function, which can be evaluated with stochastic errors. The problem consists of minimizing the sum of the agent funct...
详细信息
We consider a distributedmulti-agent network system where each agent has its own convex objective function, which can be evaluated with stochastic errors. The problem consists of minimizing the sum of the agent functions over a commonly known constraint set, but without a central coordinator and without agents sharing the explicit form of their objectives. We propose an asynchronous broadcast-based algorithm where the communications over the network are subject to random link failures. We investigate the convergence properties of the algorithm for a diminishing (random) stepsize and a constant stepsize, where each agent chooses its own stepsize independently of the other agents. Under some standard conditions on the gradient errors, we establish almost sure convergence of the method to an optimal point for diminishing stepsize. For constant stepsize, we establish some error bounds on the expected distance from the optimal point and the expected function value. We also provide numerical results.
Conducting data fusion and Command and Control (C2) in large-scale systems requires more than the presently available Peer-to-Peer (P2P) technologies provide. Resource Clustered Chord (RC-Chord) is an extension to the...
详细信息
ISBN:
(纸本)9781424452385
Conducting data fusion and Command and Control (C2) in large-scale systems requires more than the presently available Peer-to-Peer (P2P) technologies provide. Resource Clustered Chord (RC-Chord) is an extension to the Chord protocol that incorporates elements of a hierarchical peer-to-peer architecture to facilitate coalition formation algorithms in large-scale systems. Each cluster in this hierarchy represents a particular resource available for allocation, and RC-Chord provides the capabilities to locate agents of a particular resource. This approach improves upon other strategies by including support for abundant resources, or those resources that most or all agents in the system possess. This scenario exists in large-scale coalition formation problems, and applies directly to the United States Air Force's CyberCraft project. Simulations demonstrate that RC-Chord scales to systems of one million or more agents, and can be adapted to serve as a deployment environment for CyberCraft.
In this paper, a novel Robust system of multi-Dimension Education agents (RSMDEA) has been proposed. RSMDEA is based on a distributed and parallel network-computing platform with multi-agent technique and distributed ...
详细信息
In this paper, a novel Robust system of multi-Dimension Education agents (RSMDEA) has been proposed. RSMDEA is based on a distributed and parallel network-computing platform with multi-agent technique and distributed and parallel algorithms, such as co-evolutionary algorithms. RSMDEA can also be applied into multi-dimension education network platform or applications. distributed and parallel architecture is the important feature of RSMDEA. Another important feature of RSMDEA is robustness. Based on problem reduction method, robust problem of RSMDEA in ideal distributed environment has been reduced and transformed into robust problems of single agents and discrete control processes. RSMDEA can be used widely in E-learning fields, especially for today's education network with too many unsafe challenges.
The multi-agent research community is currently faced with a paradox. While promoting the use of agents as the silver bullet for various software engineering problems, it faces difficulties in presenting successful de...
详细信息
ISBN:
(纸本)1402005636
The multi-agent research community is currently faced with a paradox. While promoting the use of agents as the silver bullet for various software engineering problems, it faces difficulties in presenting successful deployments. Despite the countless multi-agent prototypes that have been developed, the number of actually deployed and in use MAS is at best very small (Gasser 2000). And as long as multi-agent frameworks continue to encounter difficulties in scaling up, it seems unlikely that this will change. This paper has two aims. First, it is an attempt to relate the scalability problem of multi-agentsystems with that of executing large numbers of concurrent threads. Second, it evaluates a CORBA/Java middle-ware layer for transparent access to distributed resources. Using such a layer, it is possible, to build multi-agentsystems that require large numbers of concurrent threads and significant memory resources.
暂无评论