Recently, there has been renewed interest in basing cryptographic primitives on weak secrets, where the only information about the secret is some non-trivial amount of (min-) entropy. From a formal point of view, such results require to upper bound the expectation of some function f(X), where X is a weak source in question. We show an elementary inequality which essentially upper bounds such 'weak expectation' by two terms, the first of which is independent of f, while the second only depends on the 'variance' of f under uniform distribution. Quite remarkably, as relatively simple corollaries of this elementary inequality, we obtain some 'unexpected' results, in several cases noticeably simplifying/improving prior techniques for the same problem. Examples include non-malleable extractors, leakage-resilient symmetric encryption, alternative to the dense model theorem, seed-dependent condensers and improved entropy loss for the leftover hash lemma. © 2013 International Association for Cryptologic Research.
CITATION STYLE
Dodis, Y., & Yu, Y. (2013). Overcoming weak expectations. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 7785 LNCS, pp. 1–22). Springer Verlag. https://doi.org/10.1007/978-3-642-36594-2_1
Mendeley helps you to discover research relevant for your work.