Azuma's Inequality, also called the Azuma-Hoeffding Inequality, is a concentration inequality bounding how far a martingale with bounded increments can stray from its starting value after a given number of steps, showing the probability of a large deviation falls off exponentially as the deviation grows. Named for Kazuoki Azuma and Wassily Hoeffding, it is a foundational tool of probability theory used throughout combinatorics and theoretical computer science to prove that quantities built from many small, weakly dependent random choices concentrate tightly around their expected value.
Facts
StatementGives a concentration result for the values of martingales that have bounded differences, bounding P(X_N - X_0 >= epsilon) by exp(-epsilon^2 / (2 sum of c_k^2)). 2 Classification
Statement Form Connections
Has Statement Form
In Branch
Source Azuma's inequality (Wikipedia)
Proved By
Source Azuma's inequality (Wikipedia)
Sources
1. Wikipedia: Azuma's inequality
WikipediaLead section, statement-form referenceQuote, Lead section, statement-form reference
In probability theory, Azuma's inequality or the Azuma-Hoeffding inequality (named after Kazuoki Azuma and Wassily Hoeffding) gives a concentration result for the values of martingales that have bounded differences.
View the Source 2. Azuma's inequality (Wikipedia)
Introduction, sentence 1
gives a concentration result for the values of martingales that have bounded differences
In Branch: Probability and Statistics, Lead sentence
In probability theory, Azuma's inequality or the Azuma-Hoeffding inequality (named after Kazuoki Azuma and Wassily Hoeffding) give
Proved By: Wassily Hoeffding, Lead paragraph
In probability theory, Azuma's inequality or the Azuma-Hoeffding inequality (named after Kazuoki Azuma and Wassily Hoeffding) gives a concentration result for the values of martingales
In Group: Probability Theory, Lead section
In probability theory, Azuma's inequality or the Azuma-Hoeffding inequality
View the SourceReader Challenges (0)
No disputes yet. Spotted an error or a better source? Open the first one.
Sign in to dispute this or suggest a correction.