Title :
An improved method of HRL based on BP neural network
Author :
Hu, Kun ; Yu, Xue-Li
Author_Institution :
Dept. of Comput. Sci. & Technol., Taiyuan Univ. of Technol., Taiyuan, China
Abstract :
An improved method of hierarchical reinforcement learning which named BMAXQ was presented in order to resolve the shortcomings of MAXQ. It amended the abstract mechanism of MAXQ and utilized the peculiarities of BP neural network. This method can make agent to find the subtasks automatically and realize parallel learning for every layer. It can be adapted to the learning task during the dynamic environment.
Keywords :
learning (artificial intelligence); neural nets; BMAXQ; HRL; backpropagetiion neural network; hierarchical reinforcement learning; learning task; parallel learning; Abstracts; Algorithm design and analysis; Biological neural networks; Heuristic algorithms; Learning; Partitioning algorithms; BP Neural Network; Hierarchical Reinforcement Learning; MAXQ; Subtask;
Conference_Titel :
Instrumentation & Measurement, Sensor Network and Automation (IMSNA), 2012 International Symposium on
Conference_Location :
Sanya
Print_ISBN :
978-1-4673-2465-6
DOI :
10.1109/MSNA.2012.6324581
