Another example that might be of interest is the use of neural networks in Monte Carlo Tree Search. In 2017 there were two papers published showing how effective the combination of NN’s and MCTS are for planning in board games.

One is an algorithm called ExIT published by Anthony et al, it showed excellent performance on a somewhat obscure board game called Hex. The authors using a typical academic lab setup to achieve this. It has a very impressive 300ish citations.

The other paper is AlphaGo, which has about 10k citations and a Netflix documentary. There are some algorithmic differences that probably make AlphaGo strictly better than ExIT but the big difference seems to mostly be that one group had like a thousand GPUs for a month.