Informatika és tudomány

  • Sir Cryalot
    #5
    alphago:
    "It's based on algorithms that have been know for several decades already. The Monte Carlo method has been invented in 1949, adapted for Tic-tac-toe in 1987, for Go in 1992, combined with tree search in 2006 and then combined with deep learning (known since 1986) through the use of many-layered feedforward neural networks (2006) that became practical recently due to the increase of power in GPUs."