Statements (19)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
gptkb:inequality
|
| gptkbp:appliesTo |
martingales
|
| gptkbp:field |
gptkb:probability_theory
statistics |
| gptkbp:generalizes |
gptkb:Hoeffding's_inequality
|
| https://www.w3.org/2000/01/rdf-schema#label |
Azuma–Hoeffding inequality
|
| gptkbp:introducedIn |
1967
|
| gptkbp:namedAfter |
gptkb:Wassily_Hoeffding
Kazuoki Azuma |
| gptkbp:provides |
tail bound
|
| gptkbp:publishedIn |
gptkb:Japanese_Journal_of_Mathematics
|
| gptkbp:sentence |
For a martingale with bounded differences, the probability that the sum deviates from its mean by t or more is at most exp(-2t^2 / sum of squared bounds).
|
| gptkbp:type |
concentration inequality
|
| gptkbp:usedIn |
gptkb:machine_learning
theoretical computer science combinatorics random processes |
| gptkbp:bfsParent |
gptkb:Ville's_inequality
|
| gptkbp:bfsLayer |
8
|