We co nsider the problem of change-point detection in mu ltivariate time-series\, typically the expression of a set of genes\, or the activity of a set of br ain regions over time. We adopt the framework of g raphical models to described the dependency betwee n the series. We are interested in the situation w here the graphical model is affected by abrupt cha nges throughout time. In the above examples\, such changes correspond to gene or brain region rewiri ng.

We demonstrate that it is poss ible to perform exact Bayesian inference whenever one considers a simple class of undirected graphs called spanning trees as possible structures. We a re then able to integrate on both the graph and se gmentation spaces at the same time by combining cl assical dynamic programming with algebraic results pertaining to spanning trees. In particular\, we show that quantities such as posterior distributio ns for change-points or posterior edge probabiliti es over time can efficiently be obtained.

We illustrate our results on both synthetic and experimental data arising from molec ular biology and neuroscience. \; LOCATION:Seminar Room 1\, Newton Institute CONTACT:INI IT END:VEVENT END:VCALENDAR