Research on coalition formation usually assumes the values of potential coalitions to be known with certainty. Furthermore, settings in which agents lack sufficient knowledge of the capabilities of potential partners is rarely, if ever, touched upon. We remove these often unrealistic assumptions and propose a model that utilizes Bayesian (multiagent) reinforcement learning in a way that enables coalition participants to reduce their uncertainty regarding coalitional values and the capabilities of others. In addition, we introduce the Bayesian Core, a new stability concept for coalition formation under uncertainty. Preliminary experimental evidence demonstrates the effectiveness of our approach. 1.

Abstract. Within the context of agent-to-agent purchase negotiations, a problem that has received little attention is that of identifying negotiation opponents in situations where the consequences of conflict and the ability to access resources dynamically vary. Such dynamism poses a number of problems that make it difficult to automate the identification of appropriate opponents. To that end, this paper describes a motivation-based opponent selection mechanism used by a buyer-agent to evaluate and select between an already identified set of seller-agents. Sellers are evaluated in terms of the amount of conflict they are expected to bring to a negotiation and the expected amount of cost a negotiation with them will entail. The mechanism allows trade-offs to be made between conflict and cost minimisation, and experimental results show the effectiveness of the approach. 1

Abstract. If we are to enable agents to handle increasingly greater levels of complexity, it is necessary to equip them with mechanisms that support greater degrees of autonomy. This is especially the case when it comes to agent-to-agent interaction which, in systems of selfish agents, often follows the format of negotiation. Within this context, a problem which has hitherto received little attention is that of identifying appropriate negotiation opponents. Furthermore, the problem is particularly difficult in dynamic systems where the need to negotiate over issues and the evaluation of resources may change over time. Such dynamics demand high degrees of autonomy from agents so that such factors can be handled at run-time and without the aid of human controllers. To that end, this paper draws inspiration from biological organisms and theories of motivation, and describes a motivation-based architecture comprising a number of motivation-based classification and selection mechanisms used to evaluate and select between negotiation opponents. Opponents are evaluated in terms of the likely issues they will want to negotiate over and the amount of conflict this might entail. Additionally, the expected cost of a negotiation with an opponent is examined in relation to the agent’s current motivational evaluation of its resources. The mechanisms allow prioritisation between each method of evaluation dependent upon motivational needs. Some preliminary evaluation of the model is also presented. 1

Sequential decision making under uncertainty is always a challenge for autonomous agents populating a multiagent environment, since their behaviour is inevitably influenced by the be-haviour of others. Further, agents have to constantly struggle to find the right balance between exploiting current information regarding the environment and the rest of its inhabitants, and ex-ploring so that they acquire additional information. Moreover, they need to profitably trade off short-term rewards with anticipated long-term ones, while learning through interaction about the environment and others—employing techniques from reinforcement learning (RL), a fun-damental area of study within artificial intelligence (AI). Coalition formation is a problem of great interest within game theory and AI, allowing autonomous individually rational agents to form stable or transient teams (or coalitions) to tackle an underlying task. Agents participating in realistic scenarios of repeated coalition formation under uncertainty face the issues identified above, and need to bargain to succesfully negotiate the terms of their participation in coalitions—often having to compromise individual with team welfare effectively. In this thesis, we provide theoretical and algorithmic tools to accommodate sequential de-

Abstract not found