regret bounds

Regret bounds provide theoretical guarantees on the performance difference between an algorithm's decisions and those of the best possible decision in hindsight, useful in online learning scenarios.

13 papers