Project
Neural Network Quantization Experiments
keywords: squeezeNet, resNet 18/34/50, model quantizaton, pytorch
date: Sep.2018
Omega is a player against AI board game based on Alpha-Beta Search framework. The board consists of fixed number of hexagonal grids, which depends on the initial setting of the board size. In the right figure, the board size is 5, it means the distance between center and every outside hex is 5. White and Black stones are allocated to each player. The stones with the same colour are marked as a group if they are connected in the board. Furthermore, the score is calculated by multiplying the members in each group. For example, white player has 3 separate groups with 3, 7 and 5 white stones respectively, hence his score is 3*7*5 = 105.
Each round has two turns: player's turn and AI's turn. Each turn player(AI) needs to firstly place his own stone and then put another stone for his rival. The game will be ended until there is no enough space for an entire round. The player with higher score is the winner.