News for hackers. Stuff that matters.

A comment on AlphaGo's next move. Parent comment

Re:AlphaGo's next move by yeukhon on Saturday May 27, 2017 @09:06AM (#14429903)
Most importanly, it is almost inevitable they must run on a fam of computers basically mean it becomes a service. Can we ever create a robot who can self-learn but with the super brain power locally without having to call a service for an answer?
[ Parent | on HN ]
Re:AlphaGo's next move by skybrian on Saturday May 27, 2017 @12:23PM (#14430711)
TPU's are available for rent. Also, apparently AlphaGo became 10x more efficient in a year. Who's to say other Go engines won't continue improving efficiency?

Between software and hardware improvements I think it's likely that we'll see very strong Go engines on desktop computers in a few years.

[ Parent | on HN ]
Re:AlphaGo's next move by wiz21c on Saturday May 27, 2017 @09:35AM (#14429991)
This raises a question for me : is there currently some public infrastructure that can rival with AlphaGo ?
[ Parent | on HN ]
Re:AlphaGo's next move by yeukhon on Saturday May 27, 2017 @09:37AM (#14429997)
Public means API? AlphaGo is very specialized in solving Go game. Google, Amazon and IBM have services for various services like image recongition and speech recongition. Startups like Clarifi also exists in that space.

The closest to a generalized AI service would probably be Watson from IBM (but I don't have experiment with it sadly so I am not sure about the usage experience).

[ Parent | on HN ]
Re:AlphaGo's next move by wiz21c on Saturday May 27, 2017 @09:56AM (#14430062)
Public in the sense of non proprietary, in this case most likely universities.

My question was more : Google probably based its tools on existing tools, and those most likely come from universities (research paper, computer infrastructures, etc.). So what are those tools, where are they ?

[ Parent | on HN ]
Re:AlphaGo's next move by Choco31415 on Saturday May 27, 2017 @12:12PM (#14430660)
Google originally used Torch7 as their ML library, but shifted to use TensorFlow in April 2016. TensorFlow is written by Google itself.

AlphaGo itself uses a method that combines Monte Carlo tree search with value and policy NN. All NN used are concurrent NN. The specifics are in a paper by David Silver et al: https://dvc0t0mx8dl84.cloudfront.net/wp-content/uploads/2016...

[ Parent | on HN ]
Re:AlphaGo's next move by nl on Saturday May 27, 2017 @09:48AM (#14430033)
> The closest to a generalized AI service would probably be Watson from IBM

Ha ha ha!

[ Parent | on HN ]
Re:AlphaGo's next move by doktrin on Saturday May 27, 2017 @10:01AM (#14430080)
>> The closest to a generalized AI service would probably be Watson from IBM

> Ha ha ha!

can you expand on that for those of us who don't know much about the field?

[ Parent | on HN ]
Re:AlphaGo's next move by yeukhon on Saturday May 27, 2017 @10:27AM (#14430186)
That's not very nice.
[ Parent | on HN ]
Re:AlphaGo's next move by nl on Saturday May 27, 2017 @09:56PM (#14432787)
Sorry, I honestly thought you were joking.

'Watson' as it is sold today is a mishmash of random, separate services. You can get the same from Google, MS or multiple other places.

'Watson' the jeopardy winning thing was an ensemble of search and rule-based NLP techniques.

Neither are much like general AI.

[ Parent | on HN ]