Name
Abstract | ||
|---|---|---|
Monte Carlo tree search algorithms, such as UCT, select the best-root-child as a result of an iterative search process consistent with path dependency. Recent work has provided parallel methods that make the search process faster. However, these methods violate the path-dependent nature of the sequential UCT process. Here, a more rapid search thus results in a higher search overhead. The cost thereof is a lower time efficiency. The concept of virtual loss is proposed to compensate for this cost. In this paper, we study the role of virtual loss. Therefore, we conduct an empirical analysis of two methods for lock-free tree parallelization, viz. one without virtual loss and one with the virtual loss. We use the UCT algorithm in the High Energy Physics domain. In particular, we methodologically evaluate the performance of the both methods for a broad set of configurations regarding search overhead and time efficiency. The results show that using virtual loss for lock-free tree parallelization still degrades the performance of the algorithm. Contrary to what we aimed at. |
Year | DOI | Venue |
|---|---|---|
2017 | 10.5220/0006205806480652 | ICAART: PROCEEDINGS OF THE 9TH INTERNATIONAL CONFERENCE ON AGENTS AND ARTIFICIAL INTELLIGENCE, VOL 2 |
Keywords | Field | DocType |
MCTS,Virtual Loss,Tree Parallelization,Search Overhead,Exploitation-exploration Trade-off | Computer science,Artificial intelligence,Machine learning | Conference |
Citations | PageRank | References |
0 | 0.34 | 2 |
Authors | ||
4 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| s ali mirsoleimani | 1 | 2 | 2.74 |
| Aske Plaat | 2 | 524 | 72.18 |
| H. Jaap van den Herik | 3 | 861 | 137.51 |
| Jos Vermaseren | 4 | 5 | 4.15 |