DeepMind AI AlphaStar Beats Starcraft II Pros TLO and MaNa

DeepMind, a Google owned company has made an AI called AlphaStar that once was challenging Go Players has recently challenged Starcraft II pro-players. These matches started taking places in December. The pro-stars took 10 consecutive loose in the series of 5 matches. Only win was when MaNa took over AI in a live streamed match that was streamed by Blizzard and DeepMind.

The match was played on the map Catalyst on a slightly outdated version of Starcraft II. This version was compatible with enabling AI Research. Before that match TLO said that he is confident that he will defeat the AI in all matches. But AI manager AlphaStar gave him surprise by coming with new strategy each time. StarCraft twitted about this victory of machine over the human as:

This is it, ladies and gentlemen! For the first time in history a pro concedes to an Artificial Intelligence! GG WP @DeepMindAI! #AlphaStarhttps://t.co/5VE3QQNqiw pic.twitter.com/0iQKT13dEA

— StarCraft (@StarCraft) January 24, 2019

10-1. @Liquid_Mana walks tall today as his harass allows him to overcome #AlphaStar!

Impressive display by @DeepMindAI, the game was absolutely F-A-N-T-A-S-T-I-C!

Did you miss the action? Catch up with the VOD here: https://t.co/zyajanP4yp pic.twitter.com/M7zIins2XV

— StarCraft (@StarCraft) January 24, 2019

There were some advantage to AlphaStar who beat the star players. First is, match used Protoss class of units that is not TLO’s style. Second is that it view whole map zoomed out. Normally player have to switch between different part of the map that takes time to see and process the data. Where AI as it is watching all zoomed out it can a lot of data of different places in single time. Some things that we were better at were action per minute and reaction time. AI’s reaction time is higher than the normal person and it perform few less action in a minute than a normal or average player.

This allowed AI to take more efficient decision while playing the game. The AI has built an in-depth training system that allows it to

The AlphaStar League ran for one week, with each of the matches producing new information that helped to refine the AI’s strategy. Over the course of that week, AlphaStar played the equivalent of 200 years worth of StarCraft II. By the end of the league session, DeepMind selected five individual agents that it determined had the least exploitable strategy and had the best chance to win. It tossed those five agents at TLO and pulled off a five-game sweep.

————————————————————————————————————————

Read More Hollywood News .

Read More Tech News .