https://github.com/liuruoze/betastar

(SSCAIT'2019) BetaStar is a StarCraft AI, written by a team at Nanjing University. It is written for StarCraft I. SSCAIT=Student StarCraft AI Tournament.
https://github.com/liuruoze/betastar

competition-code cplusplus deep-neural-networks machine-learning python scripted-agent starcraft starcraft-bot starcraft-broodwar supervised-learning

Last synced: about 5 hours ago
JSON representation

(SSCAIT'2019) BetaStar is a StarCraft AI, written by a team at Nanjing University. It is written for StarCraft I. SSCAIT=Student StarCraft AI Tournament.

• Host: GitHub
• URL: https://github.com/liuruoze/betastar
• Owner: liuruoze
• License: apache-2.0
• Created: 2021-09-03T07:22:37.000Z (about 3 years ago)
• Default Branch: main
• Last Pushed: 2021-09-03T07:26:14.000Z (about 3 years ago)
• Last Synced: 2023-10-20T21:17:59.660Z (about 1 year ago)
• Topics: competition-code, cplusplus, deep-neural-networks, machine-learning, python, scripted-agent, starcraft, starcraft-bot, starcraft-broodwar, supervised-learning
• Language: C++
• Homepage:
• Size: 3.86 MB
• Stars: 5
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# BetaStar

BetaStar is a StarCraft AI, written by a team at Nanjing University. It is written for StarCraft I. Though It contains some machine learning modules, its main strength comes from heuristic rules and human knowledge.

* It achieves the 9th in the CoG 2019 tournament (formerly CIG).

* It achieves the 5th in the SSCAIT ladder (2019-09-07), shown as below: 

![SSCAIT](doc/result_in_2019-09-07.png)

## Facts

BetaStar is based on CSE(AIIDE 2018), improving the codes and strategies against all three races, and using a new machine learning module.

## Strategies

* It exploits Carriers and Shuttles against the mechanized force of Terran.

* It utilizes High Templars against the biochemistry force of Terran.

* It improves Dark Templars against Protoss.

* It perfects Dragoons and Zealots against Zerg.

## Architectures

* It uses Eigen as a computing library and MiniDNN as a machine learning library.

* It provides all training codes(in C++) and data processing code(in Python).

## Starcraft/BWAPI-related 3rd-party Libraries Used

* BWEM

* BWEB

## Team members

Ruo-Ze Liu, Yuntao Ma, Haifeng Guo, Yuzhou Wu, Yuanhao Zheng, Zitai Xiao, Yang Yu, Tong Lu.