The PhD defence will be in Abels utsikt - Niels Henrik Abels hus. The host of the session will moderate the technicalities while the chair of the defence will moderate the disputation.
Ex auditorio questions: the chair of the defence will invite the audience to ask questions ex auditorio at the end of the defence.
Trial lecture
16th of June, time: 10:15 am, room 1259 (Abels utsikt).
E-Values, e-Processes, and modern sequential inference
Main research findings
Changepoint analysis is the problem of identifying points in time where a data sequence changes distribution. This thesis reports novel statistical methods for identifying such distributional shifts, both for retrospective data sets and for real-time streaming data. The methods are designed with two goals in mind, namely, to have low computational costs and statistically optimal performance. For retrospective data analysis, the thesis reports two new methods, one for estimating the times at which a data sequence has a change in the mean, and another for testing whether the data have a single shift in variability or co-variability. For the real-time setting, the thesis proposes a general-purpose methodology for changepoint detection that can detect a wide range of changes as they occur. All the thesis’ proposed methods are supported by strong mathematical performance guarantees regarding both their computational and statistical properties. In addition, these performance guarantees are in many instances proven to be statistically optimal, meaning that they are unbeatable from a theoretical point of view.
Adjudication committee
- Professor Tengyao Wang, London School of Economics
- Associate Professor Haeran Cho, University of Bristol
- Professor Geir O. Storvik, University of Oslo
Supervisors
- Professor Ingrid Kristine Glad, University of Oslo
- Dr. Martin Tveten, Norwegian Computing Center
Chair of defence
Professor Ulrik Skre Fjordholm, University of Oslo
Host of the session
Professor Geir O. Storvik, University of Oslo