Deep reinforcement learning for acceptance strategy in bilateral negotiations

This paper introduces an acceptance strategy based on reinforcement learning for automated bilateral negotiation, where negotiating agents bargain on multiple issues in a variety of negotiation scenarios. Several acceptance strategies based on predefined rules have been introduced in the automated negotiation literature. Those rules mostly rely on some heuristics, which take time and/or utility into account. For some negotiation settings, an acceptance strategy solely based on a negotiation deadline might perform well; however, it might fail in another setting. Instead of following predefined acceptance rules, this paper presents an acceptance strategy that aims to learn whether to accept its opponent's offer or make a counter offer by reinforcement signals received after performing an action. In an experimental setup, it is shown that the performance of the proposed approach improves over time.