Title :
Hashing a source with an unknown probability distribution
Author :
Cachin, Christian
Author_Institution :
Lab. for Comput. Sci., MIT, Cambridge, MA, USA
Abstract :
Renyi entropy of order 2 characterizes how many almost uniform random bits can be extracted from a distribution by universal hashing using a technique known as “privacy amplification” in cryptography. We generalize this result and show that if PS is the assumed distribution of a random variable with true distribution PX, then the amount of extractable almost uniform randomness corresponds to -logP[X=S], when X and S are interpreted as independent random variables
Keywords :
cryptography; data privacy; entropy; probability; random processes; Renyi entropy; almost uniform random bits; cryptography; independent random variables; privacy amplification; source hashing; universal hashing; unknown probability distribution; Computer science; Cryptography; Entropy; Measurement uncertainty; Privacy; Probability distribution; Random variables; Smoothing methods; Upper bound;
Conference_Titel :
Information Theory, 1998. Proceedings. 1998 IEEE International Symposium on
Conference_Location :
Cambridge, MA
Print_ISBN :
0-7803-5000-6
DOI :
10.1109/ISIT.1998.708874
