3-Dimensional convolutional neural networks for predicting StarCraft Ⅱ results and extracting key game situations.

Insung Baek,Seoung Bum Kim,Jean-Christophe Nebel

doi:10.1371/journal.pone.0264550

Abstract

In real-time strategy games, players collect resources, control various units, and create strategies to win. The creation of winning strategies requires accurately analyzing previous games; therefore, it is important to be able to identify the key situations that determined the outcomes of those games. However, previous studies have mainly focused on predicting game results. In this study, we propose a methodology to predict outcomes and to identify information about the turning points that determine outcomes in StarCraft Ⅱ, one of the most popular real-time strategy games. We used replay data from StarCraft Ⅱ that is similar to video data providing continuous multiple images. First, we trained a result prediction model using 3D-residual networks (3D-ResNet) and replay data to improve prediction performance by utilizing in-game spatiotemporal information. Second, we used gradient-weighted class activation mapping to extract information defining the key situations that significantly influenced the outcomes of the game. We then proved that the proposed method outperforms by comparing 2D-residual networks (2D-ResNet) using only one time-point information and 3D-ResNet with multiple time-point information. We verified the usefulness of our methodology on a 3D-ResNet with a gradient class activation map linked to a StarCraft Ⅱ replay dataset.

Highlights

The gaming industry has grown to such an extent in popularity that one segment of it has been recognized as a new competition category, e-sports [1,2]
While previous studies of StarCraft and StarCraft II mainly focused on predicting the results of games, we propose a predictive model capable of predicting game results and of identifying the key situations that influence those results
We proposed a methodology to extract key situations in real-time strategy games such as StarCraft II

Summary

Introduction

The gaming industry has grown to such an extent in popularity that one segment of it has been recognized as a new competition category, e-sports [1,2]. Various genres such as real-time strategy games, first-person shooter games (FPS), role-playing games (RPG), and multiplayer online battle arenas (MOBA) have proven especially popular. The number of people employed in e-sports, such as professional gamers, commentators, and coaching staff, continues to increase In accompaniment with this growth, the problem of extracting key game situations has become an important research topic. Extracting key game situations means using in-game information to derive important situations based on forecasting the probability of a win or a loss

Methods

Results

Conclusion