• Aucun résultat trouvé

Nested Monte-Carlo Search

N/A
N/A
Protected

Academic year: 2021

Partager "Nested Monte-Carlo Search"

Copied!
6
0
0

Texte intégral

Loading

Figure

Figure 1: The score of a leaf is the number of moves on the leftmost path
Figure 3: Distributions of the scores for a depth 60 left move problem and different search levels
Figure 5: Mean scores of the searches with memorization of the best sequence in a real-time setting.
Figure 9: Distributions of the scores for Morpion Solitaire with memorization of the best sequence
+2

Références

Documents relatifs

Ainsi, les marais ont aujourd'hui perdu tout lien «naturel» avec leur envi- ronnement, qu'il n'est pas envisageable de restaurer, et leur avenir tient en leur

The existence of Hapalemur in a region approximately halfway between the popu- lations in the far north-west and in the west would further suggest that Hapalemur may once have had

We see that (i) with 30K sims/move, many versions play the semeai whenever it is useless, and all versions play it with significant probability, what is a disaster as in real

We show through an extensive set of experiments that MCTS is a compelling solution for a pattern mining task and that it outperforms the state-of-the- art approaches (exhaustive

Given the simulations are expensive, the problem is here considered determinis- tic (no noise in the initial state nor in the chosen action)... a) Setup of the acrobot problem.

Table 8 gives the time of the parallel algorithm for various numbers of slaves, with random slaves and various fixed playout times.. random evaluation when the fixed playout time

The main contribution of the paper is the Mosaic AutoML platform, adapting and extend- ing the Monte-Carlo Tree Search setting to tackle the structured optimization problem of

In subgroup discovery and exceptional model mining, the enhancements proposed in the re- cent works can be divided into three categories: (i) The pattern language, (ii) the model