協作多智能系統在網路環境之應用

Abstract

人工智慧近年來一直是熱門的研究領域之一，其應用廣泛，無論是日常生活可見的或龐大艱深的問題都藉著人工智慧來尋找解決辦法。隨著智能個體複雜度及數量增加，所組成的多智能系統中個體的互動以及通訊網路為系統所帶來的影響也漸趨複雜。在這篇論文中，我們設計增強式學習之自動掃地機器人，並以單一機器人之學習表現作為基準來衡量多智能系統的學習行為與其所在的通訊環境之間的關係。

Artificial intelligence has been one of the hottest research trends, developing to solve from problems happen in daily life to complicated ones. As the number of intelligent entities grows, the interactions between them become more complex, particularly when communication comes into play. In this thesis, we investigate the relationships lying behind communication and multi-agent learning systems through practical learning tasks. Using single agent implemented with reinforcement learning as a benchmark, the interplay between various communication scenarios and the learning behaviour in multi-agent systems is explored.