Imitation and Cooperation in Coupled Dynamical Recognizers
Imitation and Cooperation in Coupled Dynamical Recognizers
复制标题
耦合动态识别器中的模仿与合作
DOI：
10.1007/3-540-48304-7_73
复制
发表时间：
1999
期刊：
影响因子：
--
通讯作者：
M. Taiji
中科院分区：
文献类型：
--
作者：
T. Ikegami;M. Taiji
提出了一种耦合动态识别器模型，用于模拟智能游戏玩家，使其能够模仿其他玩家的行为。一种称为动态识别器的递归神经网络被用作另一个玩家的内部模型来模仿行为。研究了Rashevskyan博弈，其中每个玩家沿着沿着一个单独的空间轴移动，以占据对另一个玩家有利的位置。虽然球员是自我中心的原则，它表明，一些利他行为将作为一个动态吸引相执行。利他行为不再是通过不断地将对手玩家仅仅建模为一个针锋相对的达特玩家来实现的。相反，参与者必须动态地改变他们的模仿模型，以实现相互合作，否则他们会走向静态的非合作纳什解。增强球员的行动模式的微小差异，被称为语用悖论，是贯穿本文的关键问题。
A coupled dynamical recognizer is proposed as a model for simulating intelligent game players, who can imitate the other player's behavior. A kind of recurrent neural network called a dynamical recognizer is used as an internal model of the other player to imitate the behavior. The Rashevskyan game is examined, where each player moves along a separate spatial axis to take an advantageous position over the other player. Though the players are egocentric in principle, it is shown that some altruistic behavior will be performed as a dynamical attractor phase. The altruistic behavior is no longer attainable by continually modeling the opponent player merely as a Tit for Tat player. Rather, players have to dynamically change their model of imitation to achieve mutual co-operation, otherwise they go to a static non-cooperative Nash solution. Enhancement of a minute difference in players' action patterns, called the pragmatic paradox, is the key issue throughout this paper.