AI-Sequencium
Different AIs for Sequencium using reinforcement learning.
DQN
A deep q-network. After training for about 100,000 games, it is still fairly awful. It needs more training and an improved setup. The basic design is inspired by https://github.com/seungeunrho/minimalRL
MCTS
A Monte Carlo Tree Search that currently works better than the DQN.