Professor Steffen Lauritzen FRS
Professor Steffen L Lauritzen FRS, MA (Msc, DPhil, DSc Copenhagen)
Professor of Statistics
Emeritus Fellow of Jesus College
steffen at stats.ox.ac.uk
+44 (0)1865 272860 (Department)
Research Interests: Bayesian networks, graphical models, causal inference, forensic genetics, structure estimation, complex stochastic systems
Graphical models are stochastic models for complex stochastic systems which exploit the notion of conditional independence to obtain modularity, flexibility, and parsimony in model specification, analysis and computation. They use mathematical graphs to express relations between entities in local terms. Graph theory and associated algorithms play a fundamental role in connection with manipulating and understanding graphical models. Their versatility have earned the models a important place in a number of applications, notably probabilistic expert systems, causal inference, forensic genetics, machine learning, and bioinformatics.
One important class of problems involves the estimation of the graphical structure from available data, generally known as model selection among statisticians and structure learning in the machine learning community. Many ad hoc methods have been developed and although their properties are better understood than they were, research is needed both to develop new methods, get a better understanding of their properties. In particular it appears that existing conceptual frameworks are not satisfactory. Much of my research effort will be devoted to this area in the coming years.
Another research area is the use of probabilistic networks in forensic science. This includes the analysis of DNA mixtures as well as issues of planning. A specific problem of interest is the reconstruction of a pedigree based on observed DNA sequences and partial information on observed individuals. This is in itself strongly related to general issues of structure estimation.
R. G. Cowell, A. P. Dawid, S. L. Lauritzen and D. J. Spiegelhalter. Probabilistic Networks and Expert Systems. Springer-Verlag, Berlin-Heidelberg-New York, 1999. 321 pp.
S. L. Lauritzen. Graphical Models. Clarendon Press, Oxford, 1996. 298 pp.
A. P. Dawid and S. L. Lauritzen. Hyper Markov laws in the statistical analysis of decomposable graphical models. The Annals of Statistics 21, 1272-1317, 1993.