English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

StarCraft, one of the most difficult esport games with long-standing historyof professional tournaments, has attracted generations of players and fans, andalso, intense attentions in artificial intelligence research. Recently,Google's DeepMind announced AlphaStar, a grandmaster level AI in StarCraft IIthat can play with humans using comparable action space and operations. In thispaper, we introduce a new AI agent, named TStarBot-X, that is trained underorders of less computations and can play competitively with expert humanplayers. TStarBot-X takes advantage of important techniques introduced inAlphaStar, and also benefits from substantial innovations including new leaguetraining methods, novel multi-agent roles, rule-guided policy search,stabilized policy improvement, lightweight neural network architecture, andimportance sampling in imitation learning, etc. We show that with orders ofless computation scale, a faithful reimplementation of AlphaStar's methods cannot succeed and the proposed techniques are necessary to ensure TStarBot-X'scompetitive performance. We reveal all technical details that are complementaryto those mentioned in AlphaStar, showing the most sensitive parts in leaguetraining, reinforcement learning and imitation learning that affect theperformance of the agents. Most importantly, this is an open-sourced study thatall codes and resources (including the trained model parameters) are publiclyaccessible via \url{https://github.com/tencent-ailab/tleague_projpage}. Weexpect this study could be beneficial for both academic and industrial futureresearch in solving complex problems like StarCraft, and also, might provide asparring partner for all StarCraft II players and other AI agents.

TStarBot-X: An Open-Sourced and Comprehensive Study for Efficient League Training in StarCraft II Full Game