#### DMCA

## Optimisation in Multi-agent Systems

### Citations

1911 |
Rate control in communication networks: shadow prices, proportional fairness and stability
- Kelly, Maulloo, et al.
- 1998
(Show Context)
Citation Context ...a ∈ A can potentially complete is denoted Ta ⊆ T. Similarly, the set of agents which can complete a task t ∈ T is denoted At ⊆ A. This problem can be conveniently represented with a factor graph (see =-=[8]-=-), which is a bipartite, undirected graph FG = {N , E}, where N is the set of nodes, such that N = VN ∪ FN , where VN is a set of variable nodes, and FN is a set of function nodes. In addition, E is a... |

297 | A Market-Oriented Programming Environment and its Application to Distributed Multicommodity Flow Problems
- Wellman
- 1994
(Show Context)
Citation Context ...d our algorithm, MIRA, to solve multi-resource allocation problems by using a generalization of Brent’s method[10] or by decomposing the market so that each market deals with only one kind of resource=-=[21]-=-. However, such extensions of MIRA are out of the scope of this paper. 4.3 MIRA algorithm Algorithm 1 shows the entire flow of the MIRA algorithm. MIRA has a distributed part and centralized part. The... |

36 |
An introduction to numerical analysis. Second edition
- Atkinson
- 1989
(Show Context)
Citation Context ...s, schedules with 3 flights, and 10 air marshals has up to 100,000 schedules and ` 100000 10 ´ joint schedules. Second, integer variables are a well-known challenge for optimization. Branch and Price =-=[1]-=- is a framework for solving very large optimization problems that combines branch and bound search with column generation to mitigate both of these problems. Column generation [6] can be viewed as a “... |

36 | Power load management as a computational market
- Ygge, Akkermans
- 1996
(Show Context)
Citation Context ... features. This reduces the search cost for similar cases (for retrieval or prior to the introduction of new cases in the case-base) to a specific set of cases with the same index as the current case =-=[22]-=- 3.2 Learning stage This stage begins with the revise phase. During this phase, the accuracy of the final solutions obtained in previous executions of the TB-CBR cycle is checked. The algorithm only c... |

31 | A probabilistic approach to optimal robust path planning with obstacles
- Blackniore, Li, et al.
- 2006
(Show Context)
Citation Context ...timize their local schedules, losing information that was needed to timely trigger policy modifications for their schedules. We encoded scenarios of the field exercise as planning problems using PDDL =-=[5]-=-. The motivation was to identify to the extent to which current automated planning technology can address complex distributed, resource-driven, and uncertain domains. Unfortunately, this proved to be ... |

29 | A probabilistic particle control approach to optimal, robust predictive control
- Blackmore
- 2006
(Show Context)
Citation Context ...ighly-coupled dependencies. Moreover we benchmark EU-DaC against other DCOP algorithms that can also provide quality guarantees: DaCSA [11] and bounded k-optimal Maximum-Gain-Message (MGM) algorithms =-=[5, 4]-=-, namely MGM-2 and MGM-3 5, by comparing their solution quality over time and the accurancy of their quality guarantees. Firstly, we describe the different network topologies and how we generate the r... |

26 | Coordinated plan management using multiagent MDPs
- Musliner, Durfee, et al.
- 2006
(Show Context)
Citation Context ...etworks, 40(2):97–111, 2002. [10] C. Kiekintveld, M. Jain, J. Tsai, J. Pita, M. Tambe, and F. Ordóñez. Computing Optimal Randomized Resource Allocations for Massive Security Games. In AAMAS-09, 2009. =-=[11]-=- M. Mavronicolas, V. Papadopoulou, A. Philippou, and P. Spirakis. A network game with attackers and a defender: A survey. In ECCS, 2006. [12] M. Mavronicolas, V. Papadopoulou, A. Philippou, and P. Spi... |

24 | An efficient motion planning algorithm for stochastic dynamic systems with constraints on probability of failure
- Ono, Williams
- 2008
(Show Context)
Citation Context ...es to be applied. A software prototype, DEFACTO, was presented and tested on a simulated environment under some simplifications (e.g., no bandwidth limitations, reliable communications, omnipresence) =-=[13]-=-. Our work also recognizes the importance of flexible frameworks to allow better human-agent interactions. The test-bed presented in this paper does not make any major simplifications, being a first s... |

16 |
A Two-Level, Protocol-Based Approach to Controlling Autonomous Oceanographic Sampling Networks
- Turner, Turner
- 2001
(Show Context)
Citation Context ...s Stackelberg games to model a variety of security domains. Game-theoretic models have been applied in a variety of security settings, such as protecting critical infrastructure [6, 11, 14]. Lawrence =-=[18]-=- applies Stackelberg games in the context of screening visitors entering the US. They have also been used for studying missile defense systems [4] and for studying the development of an adversary’s we... |

15 | Iterative risk allocation: A new approach to robust model predictive control with a joint chance constraint
- Ono, Williams
(Show Context)
Citation Context ... optimal controller then needs to generate an optimal action sequence that abides to the allocated risk at each constraint. Risk allocation for multi-agent system. Past work on risk allocation [6][13]=-=[14]-=- focused on single agent problems. In this work we extend the concept of risk allocation to multiagent systems. Figure 2 shows an example of a multi-agent system with two unmanned air vehicles (UAVs),... |

14 |
The Use of discrete moment bounds in probabilistic constrained stochastic programming models
- Prékopa
- 1999
(Show Context)
Citation Context ...that are not based on Stackelberg games to security applications. Security problems are increasingly studied using gametheoretic analysis, ranging from computer network security [17, 10] to terrorism =-=[16]-=-. Babu et al [2] have worked on modeling passenger security system at US airports using linear programming approaches, however, their objective is to classify the passengers in various groups and then... |

13 | Convex chance constrained predictive control without sampling
- Blackmore, Ono
- 2009
(Show Context)
Citation Context ...on. Branch and Price [1] is a framework for solving very large optimization problems that combines branch and bound search with column generation to mitigate both of these problems. Column generation =-=[6]-=- can be viewed as a “double oracle" algorithm [12], and is used to avoid explicitly enumerating all the variables in a large problem (in our problem, these variables represent the joint schedules). Th... |

10 | Unemployment insurance and capital accumulation
- Young
- 2004
(Show Context)
Citation Context ...the development of BDI agents following a CBR methodology [4]. Some examples of multi-agent CBR systems are: the technique called Negotiated Case Retrieval [24], the Federated Peer Learning framework =-=[23]-=-, the Collaborative CBR (CCBR) [15], the Multi-CBR (MCBR) [16], and finally, a distributed learning methodology that combines individual and cooperative learning in a MAS framework [26]. Cited above a... |

9 |
Planning with continuous resources for agent teams
- Marecki, Tambe
- 2009
(Show Context)
Citation Context ...r e-commerce applications. ISADS. [8] A. Garcia-Fornes, A. Terrasa, V. Botti, and A. Crespo. Analyzing the schedulability of hard real-time artificial intelligence systems. EAAI, pages 369–377, 1997. =-=[9]-=- A. Garvey and V. Lesser. Design-to-time real-time scheduling. IEEE Transactions on Systems, Man and Cybernetics, 23(6), 1993. [10] A. Garvey and V. Lesser. A survey of research in deliberative real-t... |

9 | Stochastic inequality constrained closed-loop model predictive control
- Hessem
- 2004
(Show Context)
Citation Context ...odel predictive control (RMPC) community. Due to the difficulty of handling the chance constraint analytically, past work used either a very conservative bound that resulted in large suboptimality [5]=-=[19]-=-, or a sample-based method [4] that is computationally inefficient. Based on the pioneer work by [16] that proposed the conservative approximation of chance constraint by decomposing it into multiple ... |

8 |
Agent-based distributed resource allocation in technical dynamic systems
- Voos
- 2006
(Show Context)
Citation Context ...es with a high probability of fulfilling the goal. On the other hand, a fearless agent will have a low confidence value. 4.2 Tests and Results The example has been implemented using the jART platform =-=[20]-=- (which is specially designed for real-time multiagent systems) and RT-Java [1] as the programming language. Once the example was implemented several simulation tests were conducted to evaluate the pr... |

7 |
Solving nonlinear simultaneous equations with a generalization of brent’s method
- Martínez
- 1980
(Show Context)
Citation Context ...h games caused by the arbitrary size and structure of possible security schedules, and the concomitant combinatorial allocations of security resources to schedules. Indeed, as shown by Korzhyk et al. =-=[10]-=-, the problem can be solved in polynomial time only if the schedules are of size 0 or 1, or if there is exactly one resource type for a schedule size of 2, and is NP-hard in general. Motivated by FAMS... |

7 | Networked distributed pomdps: A synergy of distributed constraint optimization and pomdps
- Nair, Varakantham, et al.
- 2005
(Show Context)
Citation Context ...orithm (DaCSA) [11] is a computational realization of the DaC framework, a bounded approximate DCOP algorithm in which agents coordinate by exchanging their preferred decision. However, several works =-=[3, 12]-=- have shown that communicating the utility of variables assignments instead of only the preferred assignments can lead to benefits in terms of solution quality. It is this issue that we address in thi... |

5 |
Decentralized mdps with event-driven interactions
- Becker, Lesser, et al.
- 2004
(Show Context)
Citation Context ...a Stackelberg game and solve for an optimal probability distribution over the possible deployments to ensure unpredictability. Novel solvers for classes of security games have recently been developed =-=[2, 13, 4]-=-. However, these solvers take time at least polynomial in the number of actions of both players. In our setting, every path from an entry point to a target is an attacker action, and every set of r or... |

5 |
Price Dynamics in Equilibrium Models: The Search for Equilibrium and the Emergence of Endogenous Fluctuations
- Tuinstra
- 2000
(Show Context)
Citation Context ...ake. The price must be adjusted so that the total demand (aggregate demand) becomes equal to the supply. The equilibrium price is found by an iterative process called tâtonnement or Walrasian auction =-=[17]-=- as follows: Increase the price if aggregate demand exceeds supply, Decrease the price if supply exceeds aggregate demand, Repeat until supply and demand are balanced. In classical tâtonnement, ... |

4 | The application of planning to power substation voltage control
- Bell, Coles, et al.
- 2008
(Show Context)
Citation Context ...y, so as not to enter an unsafe state during recovery. This is formally known as superstabilization [4]. In more detail, in order for an algorithm to be superstabilizing, it must be super-stabilizing =-=[3]-=-, and a pre-defined passage predicate must hold at all times. More specifically, a self-stabilizing algorithm must be distributed across a number of agents, and must be able to return those agents to ... |

3 |
An agent-based electrical power market
- Jacobo, Roure, et al.
- 2008
(Show Context)
Citation Context ...xt Wave in Computing, Optimization and Decision Technologies, pages 3–17, 2005. [6] G. Brown, M. Carlyle, J. Salmerón, and K. Wood. Defending critical infrastructure. Interfaces, 36(6):530–544, 2006. =-=[7]-=- N. Gatti. Game theoretical insights in strategic patrolling: Model and algorithm in normal-form. In ECAI, 2008. [8] C. Kiekintveld, M. Jain, J. Tsai, J. Pita, M. Tambe, and F. Ordóñez. Computing Opti... |

3 |
Risk allocation for multi-agent systems using tâtonnement
- Ono, Williams
- 2009
(Show Context)
Citation Context ...applications are using such algorithms: ARMOR and IRIS. ARMOR has been in use for over two years by Los Angeles International Airport police to generate canine-patrol and vehicle-checkpoint schedules =-=[15, 14]-=-. IRIS was recently deployed by the Federal Air Marshals Service (FAMS) to create flight schedules for air marshals [16]1. These applications are fueled by 1FAMS deploys armed air marshals on US comme... |