Potential Based Approachability

@inproceedings{PotentialBA,
  title={Potential Based Approachability},
  author={}
}
    Last lecture, we saw Blackwell's celebrated Approachability Theorem, which establishes a procedure by which a player can ensure that the average (vector) payoff in a repeated game approaches a convex set. The central idea was to construct a hyperplane separating the convex set from the point ¯ t−1 , the average loss so far. By projecting perpendicular to… CONTINUE READING