We propose a novel a framework for deriving approximations for intractable probabilistic models. This framework is based on a free energy (negative log marginal likelihood) and ca...
Collecting large consistent data sets for real world software projects is problematic. Therefore, we explore how little data are required before the predictor performance plateaus...
A common way to elude a signature-based NIDS is to transform an attack instance that the NIDS recognizes into another instance that it misses. For example, to avoid matching the a...
Abstract. Today, there exist many algorithms and protocols for constructing agregation or dissemination trees for wireless sensor networks that are optimal (for different notions o...
We consider the problem of randomness extraction from independent sources. We construct an extractor that can extract from a constant number of independent sources of length n, ea...