DMCA
Multi-objective Monte-Carlo Tree Search
| Venue: | in "Asian Conferenc on Machine Learning (ACML 2012 |
| Citations: | 2 - 1 self |
BibTeX
@INPROCEEDINGS{Wang_multi-objectivemonte-carlo,
author = {Weijia Wang and C. H. Hoi and Wray Buntine},
title = {Multi-objective Monte-Carlo Tree Search},
booktitle = {in "Asian Conferenc on Machine Learning (ACML 2012},
year = {},
pages = {507--522}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
Concerned with multi-objective reinforcement learning (MORL), this paper presents MO-MCTS, an extension of Monte-Carlo Tree Search to multi-objective sequential decision making. The known multi-objective indicator referred to as hyper-volume indicator is used to define an action selection criterion, replacing the UCB criterion in order to deal with multi-dimensional rewards. MO-MCTS is firstly compared with an existing MORL algorithm on the artificial Deep Sea Treasure problem. Then a scalability study of MO-MCTS is made on the NP-hard problem of grid scheduling, showing that the performance of MO-MCTS matches the non RL-based state of the art albeit with a higher computational cost.
Keyphrases
multi-objective monte-carlo tree search multi-objective sequential decision making scalability study known multi-objective indicator multi-dimensional reward np-hard problem morl algorithm art albeit grid scheduling action selection criterion non rl-based state computational cost multi-objective reinforcement learning artificial deep sea treasure problem hyper-volume indicator ucb criterion monte-carlo tree search
