Abstract : We propose a new point-based method for approximate planning in Dec-POMDP which outperforms the state-of-the-art approaches in terms of solution quality. It uses a heuris- tic estimation of the prior probability of be- liefs to choose a bounded number of policy trees: this choice is formulated as a combina- torial optimisation problem minimising the error induced by pruning.
https://hal.inria.fr/inria-00546912
Contributeur : François Charpillet
<>
Soumis le : mercredi 15 décembre 2010 - 10:09:47
Dernière modification le : jeudi 11 janvier 2018 - 06:19:51
Gabriel Corona, François Charpillet. Distribution over Beliefs for Memory Bounded Dec-POMDP Planning. The 26th Conference on Uncertainty in Artificial Intelligence (UAI 2010), Jul 2010, Catalina Island, California, United States. AUAI Press, 2010, 〈http://event.cwi.nl/uai2010/papers/UAI2010_0231.pdf〉. 〈inria-00546912〉
