Title :
Causal Graph Based Dynamic Optimization of Hierarchies for Factored MDPs
Author :
Hongbing Wang ; Jiancai Zhou ; Xuan Zhou
Author_Institution :
Sch. of Comput. Sci. & Eng., Southeast Univ., Nanjing, China
Abstract :
This paper presents an approach based on casual graph to optimize the task hierarchies for Hierarchical Reinforcement Learning (HRL). We conducted experiments to show that the resulting task hierarchies can improve effectiveness of reinforcement leaning.
Keywords :
dynamic programming; graph theory; learning (artificial intelligence); HRL; causal graph based dynamic optimization; factored MDP; hierarchical reinforcement learning; task hierarchy; Complex systems; casual graph; genetic programming; hierarchical reinforcement learning;
Conference_Titel :
Web Intelligence and Intelligent Agent Technology (WI-IAT), 2012 IEEE/WIC/ACM International Conferences on
Conference_Location :
Macau
Print_ISBN :
978-1-4673-6057-9
DOI :
10.1109/WI-IAT.2012.59
