Deep Learning with Shogi AI on Mac and Google Colab Chapter 12 3
expand_node()
A method that expands unexpanded nodes. For example, suppose there is a node you want to expand as shown in the figure below.
When node_expand () is executed, the index (here, x) is determined from the hash of that node, and node_hash [x] and uct_node [x] are assigned for this node. Generate the 9 parameters shown in red and save them as member variables of uct_node [x].
eval_node()
A method that inputs the aspect of the node to be evaluated and outputs the predicted probability of the policy network and the predicted winning percentage of the value network. For example, suppose there is a node you want to evaluate as shown in the figure below.
The phase information is input to the neural network to obtain the predicted probability of the policy network and the predicted win rate of the value network.
Update node information (three in red).
Recommended Posts