Generalized Strong Extractors and Deterministic Privacy Amplification

摘要: Extracting essentially uniform randomness from a somewhat random source X is crucial operation in various applications, particular cryptography where an adversary usually possesses some partial information about X. In this paper we formalize and study the most general form of extracting such cryptographic setting. Our notion strong extractors captures case catalyst neither nor independent actual extractor input. This for example important privacy amplification, key generated by Alice Bob sharing partially secret exchanging R over insecure channel accessible to Eve. Here authentication creates, Eve's viewpoint, dependence between R. We provide explicit constructions setting based on blenders. addition, give deterministic lists variables, only unknown subset variables required have amount min-entropy.