innovator

Add idea


Calendar

«    December 2018    »
MonTueWedThuFriSatSun
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
 

 

Advert

 

Payment

 

Advert

 

Authorization

Стартап

DateDate: 9-12-2018, 08:58

Engineers DeepMind presented AlphaGo Zero. She can independently learn to play any board game and beat a person, writes Science.
Alphabet's artificial intelligence division Alphabet developed the 2014 AlphaGo algorithm. Since then, the AI has won the best go players and scored 60 wins on two Chinese online platforms with the board games FoxGo and Tygem. The algorithm consists of two neural networks that mimic the work of neurons in the brain.
Now the researchers presented the new AlphaGo - the engineers created the neural network, which, unlike its predecessor, was trained not on the basis of knowledge about the strategies of the game, but made decisions from scratch - without any knowledge other than the rules.
Playing with volunteers, the system analyzed all its own and other people's moves and gradually built strategies and combinations by itself.
People learned to play chess, go, shogi and many other games for hundreds and thousands of years. AlphaZero has independently reached the pinnacle of excellence in all these games without any help from our side. My dream is to create the same system that would not only know how to play, but also solve everyday tasks, for example, create new medicines.
David Silver, DeepMind
Source: hightech.fm