This is an agent for the Hearthstone judge framework. This agent choose the most promising using Monte Carlo Tree Search with a neural network.
- Any time: This agent can be stopped at any time, and return the most promising action so far.
- Learnable: A neural network is used during the tree expansion. This neural network can be improved from training.