English

Experimental Evidence for Asymptotic Non-Optimality of Comb Adversary Strategy

Computer Science and Game Theory 2019-12-04 v1 Machine Learning

Abstract

For the problem of prediction with expert advice in the adversarial setting with finite stopping time, we give strong computer evidence that the comb strategy for k=5k=5 experts is not asymptotically optimal, thereby giving strong evidence against a conjecture of Gravin, Peres, and Sivan.

Keywords

Cite

@article{arxiv.1912.01548,
  title  = {Experimental Evidence for Asymptotic Non-Optimality of Comb Adversary Strategy},
  author = {Zachary Chase},
  journal= {arXiv preprint arXiv:1912.01548},
  year   = {2019}
}

Comments

3 pages, 1 figure