On-Line Incremental Learning in Bilateral Multi-Issue Negotiation

摘要

In this paper, we assume agents are cooperative negotiators under bounded number of negotiation messages. We implement agents who could incrementally learn from other agent's proposal during negotiation in order to speed up the negotiation process. We evaluate their performance in terms of Pareto efficiency, total utility payoffs, and number of negotiating messages. The experiments showed that negotiation learning agents could reach closer to the Pareto efficiency agreement in a much faster speed than such non-learning negotiating agents as simple random agents, rational agents, and cooperative agents.