AlphaGo Wins Match Against Human Champion

This weekend, AlphaGo won the third match in a row in the duel with 9-dan pro Lee Sedol. This is the first time, a human champion is beaten by an Artificial Intelligence in the game of Go - another huge victory for deep learning.

AlphaGo is based on Reinforcement Learning methods named "Monte Carlo rollouts" but also includes two predictive deep neural networks; one for predicting the outcome of the game given the present board situation, and another 'policy network', that is used to choose the AIs own moves but also to predict the likelihood of possible moves of the opponent.

We congratulate the team at DeepMind for this huge success!

Comments

sdf
sdf wrote:
Posted on

sdfs dsf

Leave a comment