Approximate exploitability: Learning a best response in large games — mentioned by yldedly

Wed, 05 Jul 2023 08:05:22 GMT

Another team beat AlphaZero with adversarial attacks: https://arxiv.org/abs/2004.09677

Paper and discussion history. A mention is not an endorsement.

Approximate exploitability: Learning a best response in large games — mentioned by fnbr

Wed, 29 Apr 2020 20:06:36 GMT

An additional problem with this is that they use A3C here for trading. A3C is known to not be suitable for adversarial environments (e.g. board games, like Chess). I wrote a paper that demonstrated that A3C is as exploitable as a uniform random strategy in board games (specifically, some poker variants): https://arxiv.org/abs/2004.09677 (Exploitable is a technical term that is defined in the paper; basically, it's "how much can someone who knows everything about your strategy beat you by?") So I would be very surprised if this survives contact with other traders.

Paper and discussion history. A mention is not an endorsement.

Approximate exploitability: Learning a best response in large games · HackerPapers discussions

Approximate exploitability: Learning a best response in large games — mentioned by yldedly

Approximate exploitability: Learning a best response in large games — mentioned by fnbr