## The advantages of compromising in coalition formation with incomplete information (2004)

### Cached

### Download Links

Venue: | In Proc. of AAMAS’04 |

Citations: | 17 - 0 self |

### BibTeX

@INPROCEEDINGS{Kraus04theadvantages,

author = {Sarit Kraus},

title = {The advantages of compromising in coalition formation with incomplete information},

booktitle = {In Proc. of AAMAS’04},

year = {2004},

pages = {588--595},

publisher = {ACM}

}

### Years of Citing Articles

### OpenURL

### Abstract

This paper presents protocols and strategies for coalition formation with incomplete information under time constraints. It focuses on strategies for coalition members to distribute revenues amongst themselves. Such strategies should preferably be stable, lead to a fair distribution, and maximize the social welfare of the agents. These properties are only partially supported by existing coalition formation mechanisms. In particular, stability and the maximization of social welfare are supported only in the case of complete information, and only at a high computational complexity. Recent studies on coalition formation with incomplete and uncertain information address revenue distribution in a naïve manner. In this study we specifically refer to environments with limited computational resources and incomplete information. We propose a variety of strategies for revenue distribution, including the strategy in which the agents attempt to distribute the estimated net value of a coalition equally. A variation of the equal distribution strategy in which agents compromise and agree to a payoff lower than their estimated equal share, was specifically examined. Our experimental results show that, under time constraints, the compromise strategy is stable and increases the social welfare compared to non-compromise strategies. 1.

### Citations

311 | Controlling cooperative problem solving in industrial multi-agent systems using joint intentions
- Jennings
- 1995
(Show Context)
Citation Context ..., for which we have shown via experiments, that deviation to equal distribution with compromise is beneficial. Many group formation algorithms for cooperative environments were suggested (e.g., [14], =-=[4]-=-). The revenue division is not important in such settings. Agents may have different views on the environment and tasks, and thus need to compromise in reaching a coalition formation agreement. Howeve... |

302 | Methods for task allocation via agent coalition formation
- Shehory, Kraus
- 1998
(Show Context)
Citation Context ...ration. Agents within coalitions can perform tasks that they might otherwise be unable to perform. Recognizing this, several studies have suggested mechanisms for agent coalition formation, e.g., [7] =-=[14]-=- [17]. In this paper we consider situations where self-interested agents may benefit from forming coalition when they have incomplete information about each other and there are time constraints. This ... |

221 | Coalition structure generation with worst case guarantees
- Sandholm, Larson, et al.
- 1999
(Show Context)
Citation Context ...e into consideration when deciding on whether to join a coalition. We focus on the stability of the proposed strategies, while Sandholm and Lesser focus on the agents' social welfare. Sandholm et al. =-=[11]-=- discuss the problem of identifying coalition structure that maximizes the sum of the values of coalitions. They neither discuss coalition value distribution nor the stability of forming such coalitio... |

86 |
The kernel of a cooperative game
- DAVIS, MASCHLER
- 1965
(Show Context)
Citation Context ...is distributed proportionally to the relative contribution of each member to the value of the coalition; (3) the estimated net value is distributed based on a variation of the Kernel solution concept =-=[1]-=-. We also consider variations of these strategies where coalition members compromise some of their profit computedsaccording to the Kernel, proportional, or equal distribution strategies. These strate... |

77 | Coalition formation among bounded rational agents
- Sandholm, Lesser
- 1996
(Show Context)
Citation Context ...tive environments where the value distribution is based on the Kernel. However the value of each coalition is known and the stability of the overall strategies was not considered. Sandholm and Lesser =-=[12]-=- present a coalition formation model for bounded-rational agents and a general classification of coalition games. As in [12] we also allow for varying coalitional values, however we provide the agents... |

57 |
Theories of Coalition Formation. Lawrence Erlbaum Associates
- Kahan, Rapoport
- 1984
(Show Context)
Citation Context ... to form coalition with, and (ii) the members of a coalition need to agree on the distribution of coalition gains among themselves. Most classical coalition methods address only the distribution task =-=[5]-=-. However, these two tasks are interdependent and both require a combinatorial search. A simultaneous solution of both increases the complexity further, especially in the case of incomplete informatio... |

46 | Coalition formation with uncertain heterogeneous information
- Kraus, Shehory, et al.
(Show Context)
Citation Context ...ver, these two tasks are interdependent and both require a combinatorial search. A simultaneous solution of both increases the complexity further, especially in the case of incomplete information. In =-=[8]-=- the first task was assumed for the RFP domain, however the revenue distribution method provided there suggested an equal distribution of profits among coalition members by a trusted agent. Such an ap... |

44 | Feasible Formation of Coalitions Among Autonomous Agents in Non-Super-Additive Environments
- Shehory, Kraus
- 1998
(Show Context)
Citation Context ...re, they simply follow system-imposed strategies, with no attempt to deviate. In [17] the problem of coalition formation is addressed for self-interested agents, but in superadditive environments. In =-=[13]-=-, solutions were proposed for nonsuperadditive environments where the value distribution is based on the Kernel. However the value of each coalition is known and the stability of the overall strategie... |

35 | Coalition formation through motivation and trust
- Griffiths, Luck
(Show Context)
Citation Context ... identifying coalition structure that maximizes the sum of the values of coalitions. They neither discuss coalition value distribution nor the stability of forming such coalitions. Griffiths and Luck =-=[2]-=- introduced the notion of clans, a group of agents that share similar objectives, and treat each other favorably when making decision about cooperation. They described mechanisms tosform, maintain and... |

16 |
Stability: Mechanisms for Coalition Formation in Task Oriented Domains
- Zlotkin, Rosenschein, et al.
- 1994
(Show Context)
Citation Context ...n. Agents within coalitions can perform tasks that they might otherwise be unable to perform. Recognizing this, several studies have suggested mechanisms for agent coalition formation, e.g., [7] [14] =-=[17]-=-. In this paper we consider situations where self-interested agents may benefit from forming coalition when they have incomplete information about each other and there are time constraints. This is th... |

14 |
A polynomial kernel-oriented coalition formation algorithm for rational information agents
- Klusch, Shehory
- 1996
(Show Context)
Citation Context ...laboration. Agents within coalitions can perform tasks that they might otherwise be unable to perform. Recognizing this, several studies have suggested mechanisms for agent coalition formation, e.g., =-=[7]-=- [14] [17]. In this paper we consider situations where self-interested agents may benefit from forming coalition when they have incomplete information about each other and there are time constraints. ... |

9 | On Experimental Equilibria Strategies for Selecting Sellers and Satisfying Buyers, Decision Support Systems Journal
- Goldman, Kraus, et al.
- 2003
(Show Context)
Citation Context ...ng an average profit of 50%. The actual cost of a given subtask was randomly drawn from a normal distribution with the mean cost mc, and a deviation of 2. 2 An experimental equilibrium, as defined in =-=[3]-=-, refers to an equilibrium which is measured experimentally with respect to a given set of strategies. When new strategies are introduced, the equilibrium can be re-computed. The discount factor δ was... |

5 |
Fuzzy coalition structures, Fuzzy Sets and Systems
- Mares
- 2000
(Show Context)
Citation Context ...the methods used to check the stability of a given state require that all agents hold the same beliefs about the state. More related to our work is research on fuzzy and stochastic co-operative games =-=[10]-=-. In such games agents face situations of uncertainty, including, for example, vagueness of expected coalition values and corresponding payoffs. This preliminary research attempts to find formal model... |

5 |
Incomplete Information, Incentive Compatibility and the
- Vohra
- 1999
(Show Context)
Citation Context ...e is incomplete information about coalition values. There are only few attempts to generalize the stability concepts of coalition formation, such as the core, for situations of asymmetric information =-=[16]-=-, [8]. Additionally, game theory does not provide algorithms that agents can use to form coalitions and to reach an agreement on value distribution. Thus, given a specific negotiation protocol for coa... |

4 | Bayesian Implementable of Efficient and Core Allocations. Penn. State U. Working Paper - Kwasnica - 2002 |

2 |
Issues of dynamic coalition formation among agents
- Klusch, Gerber
(Show Context)
Citation Context ...e of each possible coalition. For the problem we solve, this assumption does not hold. In real world situations, rarely do other agents know each agent’s exact value and costs of fulfilling each task =-=[6]-=-. Therefore, solutions presented in the studies discussed above are inapplicable for our problem. In particular, the methods used to check the stability of a given state require that all agents hold t... |

1 |
Convergent transfer schemes for n-person games
- Sterns
- 1968
(Show Context)
Citation Context ...l. The Kernel is non-empty for any coalition configuration. To find a revenue distribution in the Kernel, we use Stearns’ method, which converges to a Kernel point from any given revenue distribution =-=[15]-=-. The details follow: (1) Select the best coalition: rank all coalitions and choose the coalition ℭT with the highest rank (use a ranking strategy); (2) Create coalition structure S: assign agents not... |