Studying brain connectivity with EEG/MEG using time delays
Freitag, 31.10.08, 11:15-12:15, Raum 404, Eckerstr. 1
Studying brain connectivity form EEG/MEG data is\nextremely difficult because it is impossible to uniquely solve the 'inverse problem', i.e. to calculate the brain activity at specified brain locations from the measurement of external electric potentials and magnetic fields.\n\nAs a consequence, an estimate of brain activity corresponds to an incomplete demixing of channel signals. It is then unavoidable that different brain sites appear to be connected even if all brain sources were truly independent.\n\n\nTo overcome this problem we suggest to exploit the fact that the mapping of brain sources to external channels is essentially instantaneous while brain interactions involve measurable time delays. In this talk I will give an overview of our recent work based on this concept.\n\n\nSpecifically, this talk has the topics:\n\na) Role of imaginary part of cross-spectra/coherency for studying true interaction as opposed to 'artefacts of volume conduction'.\n\n\nb) Separation of interacting subsystems, which are mutually independent, using "Pairwise Interacting Sourse Analysis" (PISA).\n\n\nc) Localization of interactions.\n\n\nd) Estimating the direction of information flux robust to correlated and colored background noise.
The implied market price of weather risk
Freitag, 14.11.08, 11:15-12:15, Raum 404, Eckerstr. 1
Optimale Stopp-Probleme bei Modellunsicherheit
Freitag, 12.12.08, 11:15-12:15, Raum 404, Eckerstr. 1
We develop a theory of robust optimal stopping problems with multiple priors. Applications are given for problems in Operations Research, Microeconomics, and Finance.
Optimale Stopp-Probleme bei Modellunsicherheit
Freitag, 12.12.08, 11:15-12:15, Raum 404, Eckerstr. 1
We develop a theory of robust optimal stopping problems with multiple priors. Applications are given for problems in Operations Research, Microeconomics, and Finance.
Designs für Dosisfindungsstudien
Freitag, 6.2.09, 11:15-12:15, Raum 404, Eckerstr. 1
Understanding and properly characterizing the dose response relationship is a fundamental step in the investigation of a compound, be it a new herbicide or fertilizer, a olecular entity, an environmental toxin, or an industrial chemical.\nIn this talk we investigate the problem of deriving effcient designs for the estimation of target doses in the context of clinical dose finding. We propose methods to determine the appropriate number and actual levels of the doses to be administered to patients, as well as their relative sample size allocations.\nIt is demonstrated that the designs derived for a fixed model are rather sensitive with respect to this assumption and robust optimal designs are constructed, which take into account a set of potential dose response profiles within classes of models commonly used in drug development dose finding practice.