Search papers, labs, and topics across Lattice.
The paper introduces GRAVE2, GRAVER, and GRAVER2 algorithms to address the memory limitations of Generalized Rapid Action Value Estimation (GRAVE) in Monte-Carlo Tree Search (MCTS) for General Game Playing (GGP). These algorithms employ two-level search, node recycling, or a combination of both to reduce memory usage. The key result is that GRAVE2, GRAVER, and GRAVER2 achieve comparable playing strength to GRAVE while significantly decreasing the number of nodes stored in memory.
GRAVE2, GRAVER, and GRAVER2 let you play more general games on resource-constrained devices without sacrificing performance.
Generalized Rapid Action Value Estimation (GRAVE) has been shown to be a strong variant within the Monte-Carlo Tree Search (MCTS) family of algorithms for General Game Playing (GGP). However, its reliance on storing additional win/visit statistics at each node makes its use impractical in memory-constrained environments, thereby limiting its applicability in practice. In this paper, we introduce the GRAVE2, GRAVER and GRAVER2 algorithms, which extend GRAVE through two-level search, node recycling, and a combination of both techniques, respectively. We show that these enhancements enable a drastic reduction in the number of stored nodes while matching the playing strength of GRAVE.
