Søren Wengel Mogensen, a postdoc at the Department of Automatic Control at Lund University, will present his research on 'Learning Granger-causal graphs from partially observed multivariate time series'.
Overview
- Date:Starts 20 March 2023, 14:00Ends 20 March 2023, 15:00
- Location:Analysen, EDIT building
- Language:English
Abstract
Granger causality is an asymmetric notion of independence which describes how a system of stochastic processes evolves over time. Let A, B, and C be three subsets of coordinate processes of the stochastic system. Intuitively speaking, A is Granger-noncausal for B given C if at every point in time knowing the past of both A and C is not more informative about the present of B than knowing the past of C only.
In this talk, we will describe graphical modeling of Granger causality in partially observed time series. In this case, one can use so-called directed mixed graphs to describe the Granger causality structure of the observed coordinate processes. Several directed mixed graphs may describe the same set of Granger noncausalities and therefore it is of interest to characterize such equivalence classes. It turns out that directed mixed graphs satisfy a maximality property which allows one to construct a simple graphical representation of an entire equivalence class. This is convenient as the equivalence class can be learned from data and its graphical representation concisely describes what underlying structure could have generated the observed Granger causality structure.
We also extend the maximality result to so-called weak equivalence classes. This enables tractable learning of equivalence classes of Granger-causal graphs from data sampled from a multivariate time series, even in large networks. Moreover, this can be done in a modular fashion such that data can be seen as providing separate evidence for or against the inclusion of each edge in the output graph.
About the speaker
Søren Wengel Mogensen is a postdoc at the Department of Automatic Control at Lund University. His research focuses on graphical models and causal inference, in particular in stochastic processes. An overarching goal of Søren’s work is to develop methods for extracting causal information from partial observation of, e.g., point processes, diffusions, and time series.
This is a seminar from the DSAI seminars series held every Monday at 14:00 by the Data Science and AI division. The seminars are usually hybrid.
