A Modified Multi-size Convolution Neural Network for Winner Prediction Based on Time Serial Datasets

Researches of AI planning in Real-Time Strategy (RTS) games have been widely applied to human behavior modeling and combat simulation. Winner prediction is an important research area for AI planning, which ensures the decision accuracy. Convolution neural network has proved effective in predicting winner for RTS games. This paper focuses on modify the neural network to handle the time serial datasets. Experiments show that the modified evaluating algorithm can effectively improve the accuracy of winner prediction for time serial data in RTS games.