The Causal Interpretation of Bayesian Networks

Dr Kevin Korb

School of Computer Science & Software Engineering, Monash University

Date and time: 11.30am - 12.30pm, Monday 15th May, 2006

Venue: 10.10.03 (Building 10, Level 10, Room 3)

Chair: Xiaodong Li

Abstract:

The use of Bayesian networks for probabilistic modeling has grown substantially over the past decade.  Because of knowledge engineering difficulties, the use of Bayesian networks as causal models -- presupposed by causal discovery algorithms -- has likewise grown substantially. Despite this, numerous objections to the causal interpretation have been raised.  I counter the common (unpublised) claim that Chickering's arc reversal rule undermines a causal interpretation. The problems of faithfulness and statistical indistinguishability have given rise to a more serious scepticism about causal discovery techniques.  I will show that an interventionist account of causality, together with some plausibility assumptions, eliminate both problems.

About the speaker:

Dr. Kevin Korb is a Reader, currently with the School of Computer Science & Software Engineering, Monash University.


Seminar Organisation

Seminars are free and open to the general public. No booking is necessary. If you are interested in giving a presentation in this seminar series, or to make suggestions for speakers, please contact Xiaodong Li, the seminar co-ordinator.