
GitHub - google-deepmind/alphastar
AlphaStar is a package from DeepMind that provides the tools to train an agent to master StarCraft II offered by Blizzard Entertainment. As part of our open-sourcing efforts to drive more research interest around StarCraft II, we provide the following key offerings with this package:
AlphaStar: Mastering the real-time strategy game StarCraft II
2019年1月24日 · Now, we introduce our StarCraft II program AlphaStar, the first Artificial Intelligence to defeat a top professional player.
强化学习-星际争霸机器人AlphaStar源码分析(2)——论文 - 知乎
下图展示的是AI的训练环境,AlphaStar(图中绿色圆圈)会获取游戏中的地图、单位等信息(图中的Own units、Opponents units、Minimap)然后输出对应的动作(图中Action),例如哪个单位要移动到哪里、要建设什么单位。 动作会通过监控层(Monitoring layer)来调节频率,使得它符合人类的操作速度。 动作输入到游戏中,经过少量延迟时间(以模拟人类玩家的反应速度),AlphaStar再获取游戏信息,如此循环。 下图展示AlphaStar所取得的成果。 AlphaStar曾 …
AlphaStar (software) - Wikipedia
AlphaStar is an artificial intelligence (AI) software developed by DeepMind for playing the video game StarCraft II. It was unveiled to the public by name in January 2019. AlphaStar attained "Grandmaster" status in August 2019, a significant milestone not just for AI in video games, but arguably for the field as a whole.
AlphaStar Explained - Papers With Code
AlphaStar is a reinforcement learning agent for tackling the game of Starcraft II. It learns a policy π θ (a t ∣ s t, z) = P [a t ∣ s t, z] using a neural network for parameters θ that receives …
AlphaStar: Grandmaster level in StarCraft II using multi-agent ...
2019年10月30日 · TL;DR: AlphaStar is the first AI to reach the top league of a widely popular esport without any game restrictions. This January, a preliminary version of AlphaStar challenged two of the world's top players in StarCraft II, one of the most enduring and popular real-time strategy video games of all time.
DeepMind在线答疑:关于星际争霸AlphaStar你最需要了解的15件 …
赛后,DeepMind官方在Reddit上,回复了网友关于AlphaStar的种种疑问。 此次派出的是Oriol Vinyals和David Silver。 Oriol Vinyals 是 Google 旗下人工智能公司 DeepMind 的研究科学家,年少时,他曾是西班牙《星际争霸》游戏排名第一的电竞高手,同时参与 Gmail、Google 图像识别服务开发。 David Silver 是DeepMind首席研究员,带领DeepMind强化学习研究小组,也是AlphaGo项目的负责人。 可以说,Silver从AlphaGo诞生前起,到最强版本AlphaGo Zero,一 …
AlphaStar技术梳理 - 知乎 - 知乎专栏
PFSP:和FSP(sansi2014:MARL文章梳理:从Fictitious play到alpha-PSRO)不同,作者认为和太差(或太强)的对手打学不到什么东西,所以在FSP基础上设计一个对手采样优先级,这个技术称为Prioritized FSP
揭秘星际2人工智能AlphaStar:DeepMind科学家回应一切 | 机器之心
2019年1月26日 · 25 日凌晨,人工智能 AlphaStar 与职业玩家 MaNa 进行了一场史无前例的「人机大战」:虽然之前在内部比赛中 AI 十战十胜,但现场比赛中,MaNa 机智地戏耍了对手,为人类取得了一场胜利。 赛后,DeepMind 科学家、AlphaStar 项目的领导者 Oriol Vinyals 和 David Silver 在 Reddit 上回答了人们关心的很多问题。 与此同时,曾与人工智能交手的两位职业玩家,Liquid 战队的 TLO 与 MaNa 也作为嘉宾回答了一些有趣的问题。 例如:对于 AI 研究者来说,打星 …
AlphaStar - Zero
2021年5月1日 · AlphaStar, proposed by Vinyals et al. 2019, is the first AI agent that was rated at the Grandmaster level in the full game of StarCraft II, a real-time strategy game in which players balance high-level economic decisions with individual control of hundreds of units.