4 ms·

I would love to know where they got their training data or if there's one publicly available.

• osti 10y ago
  This is not neural net. It uses an algorithm called counterfactual regret minimization to compute the Nash equilibrium of the game, no data required.
  • sagivo 10y ago
    what about - https://arxiv.org/abs/1701.01724 https://arxiv.org/abs/1701.01724
    • osti 10y ago
      That's a new bot by university of alberta. Funnily enough, even that doesn't require actual data. It uses randomly generated ones.