Information Theory and Stochastic Inference

Information Theory and Stochastic Inference

Basic areas of competence of the team of Ulrich Sorger are probability, information, and coding theory. The main directions are decoding of error control codes and stochastic interference, where the decoding of error correcting codes can be considered as stochastic inference problem respectively the inversion of a stochastic map. Recent results show that encoding / decoding techniques exist that perform well close to theoretical limits. The team investigates these techniques and their applicability to other stochastic inference problems.
Network Traffic Modeling concerns the development of stochastic network traffic models which can help to improve performance of data transfers and network security. The aim is to use these network traffic models to derive useful conclusions from the monitored traffic concerning local congestions, localization of spam sources or denial of service (flood) attacks. Particular attention is focused on elaboration of a new approach to the detection of local network congestions based on spectral analysis of multivariate stationary processes.

Current members of the team are Foued Melakessou and Tomasz Ignatz (PhD Students) and Zdzislaw Suchanecki (senior researcher).