tailieunhanh - Multiagent-Systems 2010 Part 9

Tham khảo tài liệu 'multiagent-systems 2010 part 9', kỹ thuật - công nghệ, cơ khí - chế tạo máy phục vụ nhu cầu học tập, nghiên cứu và làm việc hiệu quả | 11 A Multi-Agent Approach to Bluffing Tshilidzi Marwala and Evan Hurwitz University of the Witwatersrand South Africa 1. Introduction The act of bluffing confounds game designers to this day. The very nature of bluffing is even open for debate adding further complication to the process of creating intelligent virtual players that can bluff and hence play realistically. Through the use of intelligent learning agents and carefully designed agent outlooks an agent can in fact learn to predict its opponents reactions based not only on its own cards but on the actions of those around it. With this wider scope of understanding an agent can in fact learn to bluff its opponents with the action representing not an illogical action as bluffing is often viewed but rather as an act of maximising returns through an effective statistical optimisation. By using a Temporal Difference-lamba TD Ấ re-inforcement learning algorithm Sutton 1988 Sutton 1989 to continuously adapt neural network agent s intelligence ability agents are shown in this chapter to be able to learn to bluff without outside prompting and even to learn to call each other s bluffs in a free competative play. While many card games involve an element of bluffing simulating and fully understanding bluffing yet remains one of the most elusive tasks presented to the game design engineer Hurwitz Marwala 2005 2007a b . The entire process of bluffing relies on performing a task that is unexpected and is thus misinterpreted by one s opponent. For this reason static rules are doomed to failure since once they become predictable they cannot be misinterpreted. In order to create an artificially intelligent agent that can bluff one must first create an agent that is capable of learning. There are many learning algorithms that have been developed and successfully implemented and these include neural networks Mohamed et al 2005 support vector machines Msiza et al 2007 and neuro-fuzzy systems Tettey Marwala 2006 . These learning