Abstract:
Statistical models allow us to predict an outcome given that we observe a set of covariates. Causal models, in addition, allow us to perform such predictions even under active perturbations of the system. And they may be helpful in another situation, too: Purely predictive methods do not perform well when the test distribution changes too much from the training distribution. Causal models may suffer less from this deficiency when the causal mechanisms stay intact. In this talk, we discuss ideas how to exploit invariance principles to learn causal models from data and to trade off between causal and predictive models to obtain models that generalize well to unseen data sets. No prior knowledge about causality is required.
About the Speaker:
Biography:
Jonas Peters is professor in statistics at the Department of Mathematical Sciences at the University of Copenhagen. He studied Mathematics at the University of Heidelberg and the University of Cambridge and obtained his PhD jointly from MPI and ETH. He is interested in inferring causal relationships from different types of data and in building statistical methods that are robust with respect to distributional shifts.
Prof. Peters won a lot of awards and Honours. COPSS Leadership Academy (2021), Guy Medal in Bronze, awarded by the Royal Statistical Society (2019), ASA Causality in Statistics Education Award (2018; with D. Janzing and B. Schölkopf), Teacher of the year at the faculty of SCIENCE, University of Copenhagen (2018), Member of the Junge Akademie (since 2016; board member since 2017), Marie Curie fellowship (2013--2015), ETH medal for an outstanding PhD thesis (2013), scholarhsip of the Studienstiftung des deutschen Volkes (2004--2008), UNWIN prize and election to scholar (Downing College, Cambridge) (2007), European Excellence Programme (DAAD), Kurt-Hahn-Trust, Hölderlin Programme (Allianz) (2006--2007), Deutsche SchülerAkademie (2001).
Prof. Peters is AE for a few top journals. Such as IEEE Transactions on Pattern Analysis and Machine Intelligence (since Jan 2021), SIAM Journal on Mathematics of Data Science (since Jan 2020), Annals of Statistics (since Jan 2019), and Journal of Causal Inference (since Jan 2021).
Zoom Meeting :https://us02web.zoom.us/j/88942756571
Meeting ID:889 4275 6571
Passcode:hYiwY7
此讲座为北京大学公共卫生学院生物统计系、北京国际数学研究中心、北京大学统计科学中心联合学术讲座。
Your participation is warmly welcomed!
欢迎扫码关注北大统计科学中心公众号,了解更多讲座信息!