Learning opening books in partially observable games: using random seeds in Phantom Go
This work addresses performance variability in randomized AIs for phantom games, which are hard for opening book learning, but is incremental as it applies existing methods to a new domain.
The paper tackled the problem of random seed variability in AI for partially observable games like Phantom Go, showing it significantly impacts performance, and applied existing algorithms to learn opening books, improving winning rates from 50% to 70% against the same AI and from near 0% to 40% against stronger opponents across board sizes.
Many artificial intelligences (AIs) are randomized. One can be lucky or unlucky with the random seed; we quantify this effect and show that, maybe contrarily to intuition, this is far from being negligible. Then, we apply two different existing algorithms for selecting good seeds and good probability distributions over seeds. This mainly leads to learning an opening book. We apply this to Phantom Go, which, as all phantom games, is hard for opening book learning. We improve the winning rate from 50% to 70% in 5x5 against the same AI, and from approximately 0% to 40% in 5x5, 7x7 and 9x9 against a stronger (learning) opponent.