AlphaGo - a historic breakthrough in AI


#1

Hello everyone,

Today AlphaGo a program designed and developed by Deepmind technologies, a subsidiary of Google and leader in AI research, has defeated Korean Grandmaster Lee-Se-dol for the third consecutive time.

“Go” is an ancient Chinese abstract strategy board game, difficult for computer programs to get trained on.

The game of Go is widely viewed as an unsolved “grand challenge” for artificial intelligence

After decades of research, computer programs have been trained on to play only on the level of human amateurs.

Deepmind technologies published a Nature paper on Mastering the game of Go with deep neural networks and tree search on January 28, 2016.

Then, AlphaGo team chose to play against the best professional “Go” game player in the last decade, Lee-Se-dol, as of today three matches are up and AlphaGo has won all the three of them. Technologists and AI enthusiasts are claiming this as a major leap in AI research.

Incase if you have missed the past matches and want to watch them,
checkout Deepmind’s Youtube channel playlist

Fourth and the final match is scheduled on 09:30AM IST, 15 March, 2016. You can watch it live in Deepmind’s Youtube channel in the link above.

And finally,

I have tried searching for source code of the program, Deepmind’s website says,

Unfortunately the code is not available to download, however click the
links above to read our Nature paper or step through the games.

Do we have any similar historic breakthrough in AI research among FOSS community?
Please share if there are any.


#2

not implemented fully yet


#3

This is a great initiative.
Hope it gets implemented successfully soon!